Package Details: llama.cpp-cuda b10091-1

Git Clone URL: https://aur.archlinux.org/llama.cpp-cuda.git (read-only, click to copy)
Package Base: llama.cpp-cuda
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: ggml, libggml, llama.cpp
Provides: llama.cpp
Submitter: txtsd
Maintainer: fabse
Last Packager: fabse
Votes: 22
Popularity: 3.04
First Submitted: 2024-10-26 20:17 (UTC)
Last Updated: 2026-07-22 22:06 (UTC)

Dependencies (19)

Required by (13)

Sources (3)

Pinned Comments

txtsd commented on 2024-10-26 20:17 (UTC) (edited on 2024-12-06 14:15 (UTC) by txtsd)

Alternate versions

llama.cpp
llama.cpp-vulkan
llama.cpp-sycl-fp16
llama.cpp-sycl-fp32
llama.cpp-cuda
llama.cpp-cuda-f16
llama.cpp-hip

Latest Comments

« First ‹ Previous 1 2 3 4 5 6 7 .. 11 Next › Last »

insonifi commented on 2026-06-18 04:19 (UTC) (edited on 2026-06-18 04:20 (UTC) by insonifi)

Adding the following cmake argument fixes the build for me:

-DGGML_CCACHE=OFF

fabse commented on 2026-06-08 09:51 (UTC)

oh hmm, did you recently upgrade to a newer cuda or? I haven't experienced this specific issue, but I once had to clear out ccache for the CI-job because a new cuda version invalidated the cache (I can see sccache was triggered in your case, so that's why I'm mentioning this)

siavoshkc commented on 2026-06-08 04:49 (UTC)

I get multiple alike errors: FAILED: [code=254] ggml/src/ggml-cuda/CMakeFiles/ggml-cuda.dir/argsort.cu.o sccache /opt/cuda/bin/nvcc -forward-unknown-to-host-compiler -ccbin=/usr/bin/g++-15 -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_FA_ALL_QUANTS -DGGML_CUDA_PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_cuda_EXPORTS -I/home/siavoshk/.cache/paru/clone/llama.cpp-cuda-git/src/llama.cpp/ggml/src/ggml-cuda/.. -I/home/siavoshk/.cache/paru/clone/llama.cpp-cuda-git/src/llama.cpp/ggml/src/../include -isystem /opt/cuda/targets/x86_64-linux/include -isystem /opt/cuda/targets/x86_64-linux/include/cccl -O3 -DNDEBUG -std=c++17 "--generate-code=arch=compute_86,code=[sm_86]" -Xcompiler=-fPIC -use_fast_math -extended-lambda -compress-mode=speed -Xcompiler "-Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wno-array-bounds -Wextra-semi -Wno-pedantic" -MD -MT ggml/src/ggml-cuda/CMakeFiles/ggml-cuda.dir/argsort.cu.o -MF ggml/src/ggml-cuda/CMakeFiles/ggml-cuda.dir/argsort.cu.o.d -x cu -c /home/siavoshk/.cache/paru/clone/llama.cpp-cuda-git/src/llama.cpp/ggml/src/ggml-cuda/argsort.cu -o ggml/src/ggml-cuda/CMakeFiles/ggml-cuda.dir/argsort.cu.o fatbinary fatal : Could not open input file 'argsort.cubin'

tuananh commented on 2026-05-24 15:01 (UTC)

Thank you @fabse

fabse commented on 2026-05-24 12:37 (UTC) (edited on 2026-05-24 13:06 (UTC) by fabse)

also, I noticed in the git-version of this package that cuda apparently doesn't compile natively by default? but it can be toggled like in this commit

I'm not using this package myself, but do ya all want that enabled?

fabse commented on 2026-05-24 12:33 (UTC)

sorry about the delay, I was waiting for this pr to be merged, which simplified building the webui, but now it should be working! lemme know if not otherwise :)

shafish commented on 2026-05-23 10:43 (UTC)

不想折腾UI问题的话直接用 llama.cpp-cuda-git 就好了

tuananh commented on 2026-05-22 15:26 (UTC)

web ui is broken. gotta add -DLLAMA_BUILD_WEBUI=ON

doaxan commented on 2026-05-20 12:30 (UTC)

Web UI is broken for b9245-1: https://github.com/ggml-org/llama.cpp/issues/23379 https://github.com/ggml-org/llama.cpp/issues/23105

fabse commented on 2026-05-07 18:51 (UTC)

oh right the config should be kept between installs, it will be included in the next release

« First ‹ Previous 1 2 3 4 5 6 7 .. 11 Next › Last »