* ggml-cuda: HIP: optimize non-saturating packed byte subtraction (`__vsubss4`) * CI: ignore 1 spilled vgpr in fattn_vec --------- Co-authored-by: Carl Philipp Klemm <carl@uvos.xyz>
* ggml-cuda: HIP: optimize non-saturating packed byte subtraction (`__vsubss4`) * CI: ignore 1 spilled vgpr in fattn_vec --------- Co-authored-by: Carl Philipp Klemm <carl@uvos.xyz>