Package Details: llama.cpp-cuda-git b9893.r0.6f8895feec-1

Git Clone URL: https://aur.archlinux.org/llama.cpp-cuda-git.git (read-only, click to copy)
Package Base: llama.cpp-cuda-git
Description: Port of Facebook's LLaMA model in C/C++ (with NVIDIA CUDA optimizations)
Upstream URL: https://github.com/ggml-org/llama.cpp
Licenses: MIT
Conflicts: llama.cpp
Provides: llama.cpp
Submitter: Bink
Maintainer: Bink
Last Packager: Bink
Votes: 6
Popularity: 2.20
First Submitted: 2026-01-08 09:17 (UTC)
Last Updated: 2026-07-07 05:16 (UTC)

Dependencies (17)

Required by (13)

Sources (3)

Latest Comments

1 2 3 4 5 Next › Last »

siavoshkc commented on 2026-07-09 19:29 (UTC)

ggml-cuda-git-1:b9940.r9.049326a000-1 and llama.cpp-cuda-git-b9862.r0.5a460dea9f-1 are in conflict

Bink commented on 2026-07-08 01:55 (UTC) (edited on 2026-07-08 01:59 (UTC) by Bink)

Thanks for bringing that issue to my attention @rubin55.

Just a few hours ago, a new commit to llama.cpp bumped the llama.cpp and ggml codebases out of alignment, causing that issue.

That could happen again in the future too, so I've pushed an update to the ggml-cuda-git dependency so it's now pulling and building GGML from upstream llama.cpp/ggml instead, which should ensure it remains aligned with this package, so long as both this package, and ggml-cuda-git are re-built together.

Bink commented on 2026-07-07 23:20 (UTC)

Certainly @rubin55.

By splitting the CUDA optimised GGML component into a separate package, multiple packages can now depend on that as a single source of CUDA optimised GGML (such as this one and whisper.cpp-cuda), solving a conflict issue.

I've replicated the issue you shared and I'm looking into it now.

rubin55 commented on 2026-07-07 20:47 (UTC)

Can I ask why this package now splits up ggml-cuda-git? I see that you're building from some other location than the vendored location previously, and I am now getting this: 

  FAILED: src/CMakeFiles/llama.dir/llama-model-loader.cpp.o
  llama-model-loader.cpp: In constructor
  'llama_model_loader::llama_model_loader(...)':
  llama-model-loader.cpp:771:18: error: 'GGML_TYPE_Q2_0' was not declared in
  this scope;
                                did you mean 'GGML_TYPE_TQ2_0'?
    771 |             case GGML_TYPE_Q2_0:    ftype = LLAMA_FTYPE_MOSTLY_Q2_0;
   break;
        |                  ^~~~~~~~~~~~~~
        |                  GGML_TYPE_TQ2_0

  ninja: build stopped: subcommand failed.
  ==> ERROR: A failure occurred in build().
      Aborting...

Bink commented on 2026-06-30 01:05 (UTC)

Thanks @ItachiSan.

The patch has now been merged upstream. I'll push an update that removes the now conflicting fix-tailwind-scanner.patch.

ItachiSan commented on 2026-06-29 20:28 (UTC)

Having a problem with the patch the last days:

makepkg
==> Making package: llama.cpp-cuda-git b9835.r1.dbdaece23d-1 (2026-06-29T22:25:16%:z)
==> Checking runtime dependencies...
==> Checking buildtime dependencies...
==> Retrieving sources...
  -> Updating llama.cpp git repo...
  -> Found llama.cpp.conf
  -> Found llama.cpp.service
  -> Found fix-tailwind-scanner.patch
==> Validating source files with sha256sums...
    llama.cpp ... Skipped
    llama.cpp.conf ... Passed
    llama.cpp.service ... Passed
    fix-tailwind-scanner.patch ... Skipped
==> Validating source files with b2sums...
    llama.cpp ... Skipped
    llama.cpp.conf ... Passed
    llama.cpp.service ... Passed
    fix-tailwind-scanner.patch ... Skipped
==> Extracting sources...
  -> Creating working copy of llama.cpp git repo...
Cloning into 'llama.cpp'...
done.
==> Starting prepare()...
The next patch would create the file tools/ui/src/.gitignore,
which already exists!  Skipping patch.
1 out of 1 hunk ignored
patching file tools/ui/sources.cmake
Hunk #1 succeeded at 13 (offset 1 line).
==> ERROR: A failure occurred in prepare().
    Aborting...

For now I've commented out the line for the patch and it seems fine.

Bink commented on 2026-06-25 23:27 (UTC) (edited on 2026-06-26 01:15 (UTC) by Bink)

@shaybox, that error is for ik-llama.cpp, not this package.

shaybox commented on 2026-06-25 23:10 (UTC)

/home/.../src/ik_llama.cpp/examples/rpc/rpc-server.cpp:266:37: error: too few arguments to function ‘ggml_backend* ggml_backend_cuda_init(int, const void*, const void*)’ 266 | backend = ggml_backend_cuda_init(device, nullptr); // init device

1 2 3 4 5 Next › Last »