PR #17576 HIP: enable WMMA-MMQ INT kernels for RDNA 3

enabled wmma instructions for most quantizations other than q2k

jiachengjason committed 55 days ago

fixed the last q2_k test case failure

jiachengjason committed 55 days ago

address comments: fix out of bound write for RDNA4, add comments after #endif

jiachengjason committed 55 days ago

clean up rebase: fix ne error in half2

jiachengjason committed 55 days ago

fix the EditorConfig CI

jiachengjason committed 55 days ago

llama.cpp HIP: enable WMMA-MMQ INT kernels for RDNA 3 #17576 Merged