Package Details: llama.cpp-hip b10068-1

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/ggml-org/llama.cpp
Licenses: MIT
Conflicts: ggml, ggml-rocm, libggml, llama-cpp, llama.cpp, stable-diffusion.cpp
Provides: ggml, ggml-rocm, libggml, libggml-hip.so, llama-cpp, llama.cpp
Submitter: Orion-zhen
Maintainer: Orion-zhen
Last Packager: Orion-zhen
Votes: 0
Popularity: 0.000000
First Submitted: 2026-07-17 15:57 (UTC)
Last Updated: 2026-07-19 02:34 (UTC)

Dependencies (19)

Required by (15)

Sources (3)

Latest Comments

Orion-zhen commented on 2026-07-18 10:43 (UTC) (edited on 2026-07-18 10:44 (UTC) by Orion-zhen)

Because the official repository handles the llama.cpp package in the following way:

  1. It separates llama-cpp and ggml into two separate packages, with llama-cpp merely serving as a frontend that calls the ggml library. (Neither of these packages are updated in real time with upstream.)
  2. The ggml package now simultaneously includes three backends—cuda, rocm, and vulkan—and activates different backends based on the software packages actually installed on the system. The trade-off is that this package will be extremely large in size.

The original AUR package was removed because there was a direct replacement in the official repository. Now that the official repository has adopted a different packaging method, I can safely resurrect the package I maintained.

s-k commented on 2026-07-18 09:13 (UTC)

why were the extra packages removed?

Orion-zhen commented on 2026-07-17 15:59 (UTC)

Haha, llama-cpp-rocm & llama-cpp-vulkan were removed from extra repo. Now my AUR packages are back!