You are not logged in.
// solved
package structure was updated - wiki already reflects the current one: https://wiki.archlinux.org/title/Llama.cpp
// original
just noticed: yay is noting that llama.cpp-hip is no longer in aur - although i just updated it today
anyone know what happened?
Last edited by cryptearth (Yesterday 20:52:52)
Offline
Offline
sorry to dig this up again - but after the replacement with llama-cpp-rocm it itself got removed, too - prompting me to switch to just llama-cpp
sorry - wasn't able to find any on the lists archive - so, what's the reason for this now? is arch just dropping llama.cpp? do i have to get back to self-build from scratch (by which it could be question: could the llama.cpp-hip be re-added?)
and please don't push me to ollama-X - i don't want to full ollama bloat - i want only llama.cpp but with amd hardware acceleraton (no matter how its called)
Offline
No, it depends on system ggml which in turn has the relevant acceleration packages, i.e. you need https://archlinux.org/packages/extra/x86_64/ggml-rocm/ which then should be available as it's backend. (... which pacman would be telling you about)
Last edited by V1del (Yesterday 18:55:03)
Offline
well - seems the wiki was worked on quite heavily over the past weeks to catch up with all those changes
guess i'll have a look there
marking as solved as the wiki reflects the changes already
Offline