mirror of
https://github.com/ROCm/composable_kernel.git
synced 2026-05-03 21:21:22 +00:00
* Update for fmha_fwd qs_ks_vs pipeline * Remove _builtin_amdgcn_sched_barrier(0) * Move p_compute to p converting earlier for trying to increase vgprs re-using * Enable GetQKBlockGemm to use WarpGemm-16x16x16 for QLoadOnce==false situation * Re-add __builtin_amdgcn_sched_barrier(0) --------- Co-authored-by: Po Yen Chen <PoYen.Chen@amd.com>