Triton autotune configs for devices with warp size != 32#1534

tridao merged 1 commit into

Dao-AILab:mainDao-AILab/flash-attention:mainfrom

schung-amd:triton-ops-warp-size-64schung-amd/flash-attention:triton-ops-warp-size-64Copy head branch name to clipboard