f6f3c44a7a
Enabling hipGraph codepath for future support ( #18 )
2022-08-09 16:45:27 -06:00
9025051bbb
Fix missing error checking for AllocateBuffs due to merge ( #17 )
2022-08-09 11:04:38 -07:00
d704668bf7
Add CMake files to build & package ( #15 )
...
* Add CMake files to build & package
* Change build technique on CI
* Correct CI build command
2022-08-09 11:17:07 -06:00
9925195afc
updated alltoallV test to not have any zero values ( #12 )
...
updated alltoallV test to not have any zero values between ranks
2022-07-21 10:28:53 -06:00
0500f2f132
implementation of multi-rank support in rccl-tests.
2022-06-10 14:54:10 -04:00
5cd2374edb
create branch up-to-date with rccl-test
2022-06-10 12:41:56 -04:00
6156759a40
Print GPU's full PCI bus ID
2022-04-06 16:46:17 +00:00
47238336d9
Update include path for custom RCCL build
2022-03-31 13:18:02 -04:00
698524e42e
move to a2a api ( #9 )
2022-02-18 08:31:40 -08:00
602b745ff4
Add missing hipStreamDestroy at test exit
2021-11-16 07:50:18 -08:00
8b35847d36
Use rccl_bfloat16 class
2021-09-23 16:39:11 -07:00
dc1ad4853d
Fix divide by zero error
2021-09-22 08:43:01 -07:00
213abee002
Merge remote-tracking branch 'nccl/master' into develop
2021-09-20 14:01:22 -07:00
f773748b46
Resync with NCCL 2.11
...
New operator: mulsum
New test: gather
2021-09-17 09:02:45 -07:00
cc34c54509
Use ROCM_PATH instead of ROCM_HOME
2021-07-21 14:19:48 -07:00
2d9be62621
Merge remote-tracking branch 'nccl/master'
2021-07-15 13:54:43 -07:00
1f8f541686
Add CUDA graph support only for CUDA 11.3 and later builds
...
Fixes #90
2021-07-13 10:47:47 -07:00
9f8ddadcdf
Merge remote-tracking branch 'nccl/master' into develop
2021-07-13 08:11:44 -07:00
b9f90d12a9
Removed MPI_SUPPORT conditional compilation of average flag
2021-07-12 11:43:57 -07:00
547e119d35
Fix issues with MPI_Allreduce and multi-threaded tests
2021-07-08 16:42:40 -07:00
f476f4a17a
Merge branch 'bfloat16'
2021-07-06 10:20:32 -07:00
1dfc76eccc
Added new option to report average iteration time
2021-06-30 19:36:07 -07:00
1ae8cdc315
Resync with changes in gitilab-master code
2021-06-30 13:16:04 -07:00
9dae3d3a37
Added new tests: scatter, sendrecv, hypercube
2021-06-28 16:49:10 -07:00
e55ad3796d
Added support for CUDA graph capture/replay (-G)
2021-06-28 14:19:45 -07:00
526eacadf7
Fixed formatting for bfloat16 support
2021-06-28 10:12:34 -07:00
cde7e769c1
Add support for ncclAvg operation
2021-06-28 09:41:58 -07:00
c4de829d91
Cleanup argument error handling and messages
...
Add error checking for minbytes and maxbytes arguments
Also accept lowercase literals when parsing size arguments and print errors and usage on stderr.
2021-06-04 21:47:40 +00:00
0fccaec26f
Update mpich include path
2021-04-16 18:25:54 -04:00
e37545e491
Add support for new datatype: bfloat16
2021-03-15 17:13:35 -07:00
39086cdc0a
Revert "Allow call ncclCommAbort on Ctrl+C"
...
This reverts commit 23c374475f
2021-02-03 21:16:18 -05:00
04422e03fd
Merge remote-tracking branch 'nccl/master'
2021-02-03 21:01:24 -05:00
7677f3f608
Do not allocate memory for expected buffer if checking disabled
...
This allows the tests to be run with larger buffers
2021-01-20 17:08:40 -08:00
e5f1482efb
Add tests code that can print info and reset input/output buffers
2021-01-04 16:51:16 -05:00
ae1ce98e69
Add boot_id to the hostname hash due to collisions on Azure
...
Fixes #60
2021-01-04 11:38:45 -08:00
da67a81c8e
Use DJB2a hash algorithm in getHostHash()
2020-12-18 10:12:54 -08:00
3117033150
Add support for testing memory allocated with hipMallocManaged
2020-12-15 22:05:50 -05:00
d310466d88
Fix alltoallv test
2020-12-14 18:03:06 -05:00
e3f9e281f1
Fix mpich linking option
2020-10-19 14:43:01 -07:00
bf4a866109
Uses nullptr as send buffer for non-root ranks during scatter
2020-10-19 14:06:23 -04:00
4c0ec7347c
Fix build error
2020-09-25 18:06:09 +00:00
3f1dfacc95
Add test for alltoallv API
2020-09-24 14:24:52 -04:00
0d1940e18e
Prioritize NCCL_HOME
2020-09-24 14:18:23 -04:00
58dcd35af2
Add alltoallv test
2020-09-01 18:55:25 +00:00
3d63a84d97
Add cumask option
2020-08-21 21:34:55 +00:00
346cb16442
Change scatter and gather bandwidth calculation to match alltoall ( #7 )
2020-07-06 17:12:50 -07:00
5361dd8177
Merge remote-tracking branch 'nccl/master' into HEAD
2020-07-06 17:54:31 +00:00
afdaf59b3b
Fix typo in src/Makefile
2020-06-24 14:39:22 -07:00
b2603a2e85
Add gencode for CUDA11
2020-06-23 18:16:46 -07:00
ec1b5e22e6
Change all_gather/reduce_scatter algbw to match the documentation.
...
Fix #45 : All_gather and reduce_scatter algorithm bandwidth was
computed as time/count*(nranks-1) which is not consistent with the
way we compute it for other collectives.
This change makes algbw higher; busbw is unchanged.
2020-06-19 10:42:19 -07:00
gilbertlee-amd
Wenkai Du
Liam Wrubleski
akolliasAMD
Edgar
Wenkai Du
Ziyue Yang
David Addison
Greg Inozemtsev
Jithin Jose
Luke Yeager
Sylvain Jeaugey