Lequn Chen
@abcdabcd987
Faster and cheaper LLM inference.
加入 January 2012
674 正在關注 1.7K 粉絲
Faster than DeepEP for Decode on ConnectX-7. First viable kernel on EFA. SM-Free RDMA transfer. Support prefill. (Maybe portable to other hardware as well)
熱門用戶
