Add support for GPU int2 prefill#316

Open

huing4257 wants to merge 4 commits intomicrosoft:mainfrom

huing4257:gpu-gemm-kernel

Commits on Jul 30, 2025

Commits on Aug 6, 2025

Fuse convert to bf16 and scaling into kernel; make BLOCK_SIZE_M and BLOCK_SIZE_N adjustable
huing4257
committed