microsoft/mscclpp
1
0
Fork 0
mirror of https://github.com/microsoft/mscclpp.git synced 2026-05-19 20:40:13 +00:00
Code Issues Packages Projects Releases Wiki Activity
249 Commits 107 Branches 16 Tags
212f92cfba7744a3ded7919955ff9502f02a3dbf
Commit Graph

249 Commits

This Branch
This Branch
All Branches
Author SHA1 Message Date
Binyang Li
212f92cfba fix 2023-04-02 11:13:43 +00:00
Binyang Li
36c418a239 Merge branch 'main' into binyli/mscclpp-test 2023-04-02 08:13:48 +00:00
Saeed Maleki
bfbdaf6b05 Merge pull request #41 from microsoft/saemal/allgather_hier 
Saemal/allgather hier
 2023-04-01 20:37:43 -07:00
Saeed Maleki
5ff64d36f4 documents for allgather2 + refactoring local allgather 2023-04-02 03:36:22 +00:00
Saeed Maleki
0887cfe768 no need for remapping anymore 2023-04-02 02:35:08 +00:00
Saeed Maleki
5cf3f3c524 Merge pull request #39 from microsoft/binyli/numabindAPI 
bind proxy threads and host allocation are now done automatically and closest to the device
 2023-04-01 19:16:43 -07:00
Saeed Maleki
1731911d00 removing extra stream and destroying created ones 2023-04-02 02:07:41 +00:00
Saeed Maleki
4c6616e7b9 lint 2023-04-01 19:20:50 +00:00
Saeed Maleki
8927dd4d72 great allgather numbers with the current binding mechanism 2023-04-01 18:54:42 +00:00
Binyang Li
f02ba52b0a fix 2023-04-01 07:10:20 +00:00
Binyang Li
9c1be4c03a fix 2023-04-01 05:10:14 +00:00
Saeed Maleki
701255959e lint 2023-03-31 23:34:43 +00:00
Saeed Maleki
97dadd8d64 merged with main 2023-03-31 23:32:01 +00:00
Binyang Li
a3adc99b5c more code 2023-03-31 15:10:51 +00:00
Binyang Li
8e4edd4d83 fix lint 2023-03-31 08:32:45 +00:00
Binyang Li
af5825b474 bind numa node to communicator 2023-03-31 08:05:49 +00:00
Binyang Li
98020f5b52 update 2023-03-31 07:10:43 +00:00
Saeed Maleki
44e8760af7 allgather kernel2 2023-03-31 06:31:25 +00:00
Changho Hwang
fe1d7fee9e Bug Fix: null-termination in logging 2023-03-31 05:25:07 +00:00
Binyang Li
c7b3d0b0fd update 2023-03-31 03:58:09 +00:00
Saeed Maleki
fef0bff945 a third kernel for allgather cross-node 2023-03-30 23:24:04 +00:00
Saeed Maleki
29254439e5 Merge pull request #38 from microsoft/saemal/removing_gdrcopy 
removing gdrcopy and adding flush functionality
 2023-03-30 13:17:47 -07:00
Binyang Li
20b644b88d port more code from nccl-test 2023-03-30 09:13:28 +00:00
Changho Hwang
b58eae4037 Minor changes 2023-03-30 07:11:41 +00:00
Binyang Li
22a977e730 init 2023-03-30 06:29:38 +00:00
Saeed Maleki
e2cfd5ac83 a lot of documentation 2023-03-30 00:37:33 +00:00
Saeed Maleki
be5e422021 merged with main 2023-03-29 23:03:12 +00:00
Saeed Maleki
4be1dcc5d6 Merge pull request #37 from microsoft/saemal/removing_gdrcopy_flush 
flush capability added in this PR
 2023-03-29 15:44:30 -07:00
Saeed Maleki
629d59a9c0 bug fix -- flush doesn't need to increment the epoch 2023-03-29 22:21:24 +00:00
Saeed Maleki
debd110874 fused flush instructions 2023-03-29 22:17:02 +00:00
Saeed Maleki
42b11c5c9a fast flush 2023-03-29 20:50:01 +00:00
Saeed Maleki
d97bee6973 flush mechanism 2023-03-29 17:31:20 +00:00
Bin Wang
7880be8ee2 Fix the 2 GiB limit in allgather test. (#36) 2023-03-29 19:02:43 +08:00
Saeed Maleki
7a0962e4be Merge pull request #35 from microsoft/binyli/refactor 
Refactor mscclppProxyService
 2023-03-28 13:46:16 -07:00
Binyang Li
d725e45f13 fix 2023-03-28 14:53:08 +00:00
Binyang Li
9c633a9633 bug fix 2023-03-28 14:40:51 +00:00
Binyang Li
487030887b refactor 2023-03-28 12:22:43 +00:00
Saeed Maleki
17e144c774 a typo in p2p proxy 2023-03-28 08:07:54 +00:00
Saeed Maleki
81b18cd9f9 a bit of clean up 2023-03-28 06:08:12 +00:00
Binyang2014
62279b0063 Add mscclppSetBootstrapConnTimeout (#34) 2023-03-28 14:01:56 +08:00
Saeed Maleki
fa26bdd9fc no gdr copy anywhere in the code except for the files that are not compiled 2023-03-28 05:40:40 +00:00
Saeed Maleki
33af4bfb67 no gdr copy anywhere in the code except for the files that are not compiled 2023-03-28 05:36:31 +00:00
Saeed Maleki
d9ba953fb0 gdrcopy is not initialized 2023-03-28 04:56:06 +00:00
Saeed Maleki
e7cccbf897 both head and tail are on OK to be only used by GPU 2023-03-28 04:26:39 +00:00
Saeed Maleki
952d852256 both head and tail are on OK to be only used by GPU 2023-03-28 04:24:45 +00:00
Ziyue Yang
b234cf5012 NPKit: add DMA events and fix bandwidth calculation (#33) 2023-03-28 09:58:32 +08:00
Saeed Maleki
ea3bd90303 Merge pull request #32 from microsoft/chhwang/log-handler 
Add mscclppSetLogHandler
 2023-03-27 17:59:00 -07:00
Saeed Maleki
32c4498fb8 typo fixes 2023-03-28 00:55:41 +00:00
Saeed Maleki
75036c0f12 typo fixes 2023-03-28 00:50:59 +00:00
Saeed Maleki
5adf3e3755 typo fix 2023-03-27 23:42:43 +00:00