[None][perf] Fuse MiniMax-M3 MSA per-layer KV-cache writes into one kernel#16755

pcicotti merged 2 commits into

NVIDIA:feat/m3_with_msaNVIDIA/TensorRT-LLM:feat/m3_with_msafrom

zheyuf:perf/m3-msa-prepzheyuf/TensorRT-LLM:perf/m3-msa-prepCopy head branch name to clipboard