[None][perf] Enable tiled FP8 context FMHA for head_dim 256 on SM120 - #19825

[None][perf] Enable tiled FP8 context FMHA for head_dim 256 on SM120#19825

amukkara merged 1 commit into

NVIDIA:mainNVIDIA/TensorRT-LLM:mainfrom

amukkara:fmha-sm120-tileamukkara/TensorRT-LLM:fmha-sm120-tileCopy head branch name to clipboard