I would like to request for this famous inference engine to be added: AUR (en) - llama.cpp-vulkan
I agree! This would be a good reason for someone to move to cachy.
I agree, the packages they have now don’t include cuda or the systemd unit.
OR THE llama-server BINARY…
Wasted hours of my life, add -cachyos to the package name if its custom, its not based on any of the existing packages.
Umm it is in the repo and includes the binary you list.
Looks like llama-cpp-vulkan and llama-cpp-rocm are in the repo now, but llama-cpp-cuda is missing.