dsa(indexer_backward): opt-in SM100 sparse backward v2 — 1.16-1.92x faster (fp32-accurate d_index_k at no extra cost) - #640

dsa(indexer_backward): opt-in SM100 sparse backward v2 — 1.16-1.92x faster (fp32-accurate d_index_k at no extra cost)#640

Anerudhan merged 6 commits into

NVIDIA:developNVIDIA/cudnn-frontend:developfrom

zkyue:feat/dsa-idxbwd-sm100-v2zkyue/cudnn-frontend:feat/dsa-idxbwd-sm100-v2Copy head branch name to clipboard