RS + GEMM overlap for BF16, gfx950, w/ HipKittens + UserBuffers - #725

Open

alextmagro wants to merge 3 commits into

devfrom

hipkittens_overlap_rs

Commits on Sep 1, 2026

Reduce scatter overlap kernels for gfx950 + bf16
alextmagro
committed
392fd78

Commits on Sep 3, 2026

Address claude comments
alextmagro
committed
d83daa7
Fix build error, sentinel correctness bug and address PR comments
alextmagro
committed
1548db7