$ the-wire · showcase
ARCH PURGES GPU ACCELERATION PACKAGES FROM EXTRA
By RepoJournal · Filed · About Arch Linux · Composed from the cited sources · methodology
Four AI and speech processing packages just disappeared from the extra repository, signaling either a maintenance consolidation or upstream trouble.
The repository shed llama-cpp-rocm [1] and llama-cpp-vulkan [2] alongside their whisper-cpp counterparts [3] [4], all from extra-x86_64. That's the entire GPU-accelerated variant stack for two critical inference tools, which suggests either a dependency failure across AMD and Vulkan backends or a deliberate decision to stop maintaining those specific builds. The timing matters: if you're running production on either of these packages, you need to know whether this is temporary or permanent. On the lighter side, gitit moved from staging into the main extra repository [5], which means it cleared testing and is now production-ready. All five changes landed within 24 hours, which is either coordinated maintenance or reactive pruning.
Action items
- → Check if you depend on llama-cpp-rocm, llama-cpp-vulkan, whisper-cpp-rocm, or whisper-cpp-vulkan archlinux/packages [immediate]
- → If using GPU variants, switch to the CPU-only builds or verify a workaround exists archlinux/packages [immediate]
- → Monitor the mailing list for removal notices or updates on the deprecation archlinux/packages [monitor]
References
- [1] remove llama-cpp-rocm from extra-x86_64 archlinux/state ↗
- [2] remove llama-cpp-vulkan from extra-x86_64 archlinux/state ↗
- [3] remove whisper-cpp-vulkan from extra-x86_64 archlinux/state ↗
- [4] remove whisper-cpp-rocm from extra-x86_64 archlinux/state ↗
- [5] move gitit from extra-staging-x86_64 to extra-x86_64 archlinux/state ↗
