| Git Clone URL: | https://aur.archlinux.org/llama.cpp-hip.git (read-only, click to copy) |
|---|---|
| Package Base: | llama.cpp-hip |
| Description: | Port of Facebook's LLaMA model in C/C++ (with AMD ROCm optimizations) |
| Upstream URL: | https://github.com/ggerganov/llama.cpp |
| Licenses: | MIT |
| Conflicts: | llama.cpp |
| Provides: | llama.cpp |
| Submitter: | txtsd |
| Maintainer: | txtsd |
| Last Packager: | txtsd |
| Votes: | 6 |
| Popularity: | 1.25 |
| First Submitted: | 2024-10-26 19:54 (UTC) |
| Last Updated: | 2025-06-11 11:33 (UTC) |
I'm getting an error when installing
error: unresolvable package conflicts detected error: failed to prepare transaction (conflicting dependencies) :: blas-openblas-0.3.29-2 and blas-3.12.1-2.1 are in conflict
But blas is required by the following packages: cblas lapack lib32-blas proton-cachyos suitesparse
I can't remove it because proton-cachyos is using it, and also it seems blas is higher than blas-openblas.
@txtsd Please set CMAKE_BUILD_TYPE to Release. Otherwise this package is pretty useless. ;)
@hwkiller Thanks for suggestion, changing build type increased speed dramatically, from unusable to 20+
I would highly suggest that you change the -DCMAKE_BUILD_TYPE=None to -DCMAKE_BUILD_TYPE=Release
By default, it produces a Debug build, and it is insanely slow on AMDGPUs. I rebuilt this using -DCMAKE_BUILD_TYPE=Release, and it went from seconds per token to 30 tokens / second.
@KritaFrost Thanks, I pushed an update.
You probably want to change -DGGML_HIPBLAS=ON to just DGGML_HIP=ON. Here link to reddit post where solution was found on why ROCm doesn't work in current build
My bad—it just doesn’t like mold linker. Switched to ldd, works great. Thank you, problem solved.
@Althorion It builds for me. Please try a clean build.
The linker dumps core now:
[ 33%] Linking CXX executable ../../bin/llama-run
clang++: error: unable to execute command: Aborted (core dumped)
clang++: error: linker command failed due to signal (use -v to see invocation)
make[2]: *** [examples/rpc/CMakeFiles/rpc-server.dir/build.make:110: bin/rpc-server] Error 1
make[1]: *** [CMakeFiles/Makefile2:4638: examples/rpc/CMakeFiles/rpc-server.dir/all] Error 2
@Althorion Thanks for reporting! I've pushed a fix. Please let me know if it works as expected.
Pinned Comments
txtsd commented on 2024-10-26 20:15 (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