7 Commits

Author	SHA1	Message	Date
Ziyue Yang	73c6d51454	fix min, max and avg (#745) ... [ROCm/rccl commit: 11676267b5]	2023-05-18 11:02:59 -07:00
Wen-Heng (Jack) Chung	932822fc46	Address review feedbacks and make the flag be disabled by default. ... [ROCm/rccl commit: ca4a1dfd67]	2023-05-17 17:50:25 +00:00
Wen-Heng (Jack) Chung	a7fbe68535	Skip doneEvent inside MSCCL by default. ... Added a RCCL_MSCCL_ENABLE_DONE_EVENT env var, set it be 0 by default. The env var is to control whether to use doneEvent when invoking MSCCL kernels. Skipping doneEvent would cause the firmware to skip L2 cache flush, resulting in overall performance improvement. [ROCm/rccl commit: 12dba425de]	2023-05-17 16:49:42 +00:00
Wenkai Du	1af52edd6a	msccl: print stack and memory usage (#723) ... * msccl: print stack and memory usage * Update number of kernels calculation [ROCm/rccl commit: 4b09ffba43]	2023-04-14 14:59:03 -07:00
Ziyue Yang	aa605b0989	fix msccl stream usage (#717) ... [ROCm/rccl commit: c8e33b1232]	2023-03-24 10:59:36 -07:00
Ziyue Yang	f7f669e7f0	MSCCL: Improve executor and integrate scheduler (#694) ... * MSCCL: improve executor and add scheduler for testing * Use external scheduler * Fix cmake error * Address comments * Fix thread safe issue * Make MSCCL lifecycle APIs thread safe * Make MSCCL internal scheduler aware of topology hint * Revise error message [ROCm/rccl commit: e3b2342f39]	2023-03-14 14:34:25 -07:00
Ziyue Yang	431c7d2cb9	Add MSCCL Support (#658) ... * Add MSCCL support * Add alignment and message size checking * Fix nRanks checking, in-place and out-of-place tests and group call handling * Fix hipGraph unit test * Change MSCCL init warning to INFO * Revise license info [ROCm/rccl commit: adafc0f759]	2022-12-12 15:51:04 -08:00