 Ziyue Yang
|
e3d45f9de4
|
Improve MSCCL algorithms (#1023)
[ROCm/rccl commit: 0a53077c9c]
|
2024-01-03 14:51:34 -08:00 |
|
|
Ziyue Yang
|
62299668bd
|
Tune MSCCL all-reduce algorithm (#1009)
[ROCm/rccl commit: bb144dcd50]
|
2023-12-08 17:47:02 -06:00 |
|
 Wen-Heng (Jack) Chung
|
0266febb31
|
Let 320KB message size uses LL protocol. (#1006)
[ROCm/rccl commit: 8e8323252a]
|
2023-12-06 18:14:31 -06:00 |
|
|
Ziyue Yang
|
cef45b8311
|
Fix mscclAlgoHandle not initialized issue (#995)
[ROCm/rccl commit: e44e112a17]
|
2023-12-01 07:58:01 -08:00 |
|
|
Ziyue Yang
|
f0c47d085e
|
Move MSCCL algorithm loading to initialization to workaround HIP graph conflict (#982)
* MSCCL: pre-specify channels and pre-load algorithms
* add mutex
* fix bug
* clean include
* disable all-gathers temporarily
[ROCm/rccl commit: 4bb0b4a380]
|
2023-11-30 09:47:20 -08:00 |
|
|
Ziyue Yang
|
2351578d5b
|
Optimize MSCCL all-gather algorithms for gfx942 (#964)
[ROCm/rccl commit: 7ae95db5b8]
|
2023-11-15 08:18:59 -08:00 |
|
 akolliasAMD
|
691df735a3
|
Revert "Introduce allgather for MSCCL on 8 sockets up to 320KB. (#931)" (#939)
This reverts commit 769f00db5c.
[ROCm/rccl commit: 9f02ee8dea]
|
2023-10-30 23:52:58 -06:00 |
|
|
Wen-Heng (Jack) Chung
|
769f00db5c
|
Introduce allgather for MSCCL on 8 sockets up to 320KB. (#931)
[ROCm/rccl commit: bfb8642450]
|
2023-10-24 18:41:12 -05:00 |
|
|
Wen-Heng (Jack) Chung
|
89a8493ef8
|
Introduce allgather MSCCL XML specification for MI250X up to 320KB. (#930)
[ROCm/rccl commit: 3f9ffe4788]
|
2023-10-24 18:35:55 -05:00 |
|
|
Wen-Heng (Jack) Chung
|
fc2a13c077
|
Introduce 1-shot allreduce for MI250X Hayabusa. (#929)
[ROCm/rccl commit: 72d5fbddfd]
|
2023-10-24 16:31:18 -05:00 |
|
|
Wen-Heng (Jack) Chung
|
49e52e7269
|
Introduce 1pass allreduce. Tailor it for very small message sizes <= 20KB. (#919)
[ROCm/rccl commit: 341926c60a]
|
2023-10-16 16:31:08 -05:00 |
|
 Wenkai Du
|
af04103d72
|
Add MSCCL xml files (#861)
[ROCm/rccl commit: aeca1af374]
|
2023-08-23 14:12:34 -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 |
|