Skip to content

cuda: smaller KV tile for the 8-column MMA flash attention config (head size 256) - #308

Open
sb32445 wants to merge 1 commit into
PrismML-Eng:prismfrom
sb32445:pr/fattn-mma-tile
Open

cuda: smaller KV tile for the 8-column MMA flash attention config (head size 256)#308
sb32445 wants to merge 1 commit into
PrismML-Eng:prismfrom
sb32445:pr/fattn-mma-tile

Commits

  1. Commits on Oct 4, 2026