Skip to content

CUDA: vectorize FP16 attention input preprocessing - #2436

Open
john-sp wants to merge 1 commit into
LeelaChessZero:masterfrom
john-sp:cuda-pr2-preprocess-fp16x8
Open

CUDA: vectorize FP16 attention input preprocessing#2436
john-sp wants to merge 1 commit into
LeelaChessZero:masterfrom
john-sp:cuda-pr2-preprocess-fp16x8

Commits

Commits on Aug 1, 2026