AMD ROCm support with HIPBLAS
xangelix opened this issue · comments
ggerganov/llama.cpp#1087 was merged recently and should make its way into GGML shortly. I think (hopefully) it should require only minimal changes on this side.
related to ggerganov/ggml#472
rustformers
uses llama.cpp
as it's ggml source, feel free to create an PR including this change, seams like you only need to adjust the build.rs
of the ggml-sys
create. I wont be able to test this as i don't own a amd gpu.