3389838def
Rework network proxy profiling
...
[ROCm/rccl commit: 7484e53ff7
2020-06-13 03:13:58 +00:00
a8952356d1
Reduce RCCL kernel count as we don't pass first coll in argument
...
[ROCm/rccl commit: b257676f30
2020-06-12 21:30:04 +00:00
a90fac8579
Sender rank's opCount maybe ahead by one if it finishes earlier
...
[ROCm/rccl commit: a6d621176c
2020-06-12 03:39:45 +00:00
d1f8fdc3a8
gtest: add scatter, gather and all to all unit tests
...
[ROCm/rccl commit: fee1a20b74
2020-06-09 17:44:15 -07:00
e7dd92dc87
Log NUMA node of RDMA host buffer allocation
...
[ROCm/rccl commit: e98891d039
2020-06-09 17:44:15 -07:00
846e1a4427
Add network proxy profiling support
...
[ROCm/rccl commit: 812543104d
2020-06-09 17:44:15 -07:00
bc6840f937
Calculate and use total wait cycles for RCCL profiling
...
[ROCm/rccl commit: c9aa11928a
2020-06-09 17:44:15 -07:00
15cd2aff3c
Add gather, scatter and alltoall collectives
...
Introducing 3 new APIs:
ncclResult_t ncclGather(const void* sendbuff, void* recvbuff, size_t sendcount,
ncclDataType_t datatype, int root, ncclComm_t comm, hipStream_t stream);
ncclResult_t ncclScatter(const void* sendbuff, void* recvbuff,
size_t recvcount, ncclDataType_t datatype, int root, ncclComm_t comm,
hipStream_t stream);
ncclResult_t ncclAllToAll(const void* sendbuff, void* recvbuff, size_t count,
ncclDataType_t datatype, ncclComm_t comm, hipStream_t stream);
Only out of place operation is supported.
Preprocessor symbol RCCL_GATHER_SCATTER=1 indicates API availibility.
By default the APIs launche RCCL kernel implementation, which can be disabled by
RCCL_ALLTOALL_KERNEL_DISABLE=1. Then the APIs use wrapper around ncclSend and ncclRecv.
[ROCm/rccl commit: e80e29573c
2020-06-09 17:44:08 -07:00
69eb70ce43
tpol_expl: update to 2.7
...
[ROCm/rccl commit: 71ec3e09df
2020-06-09 17:40:24 -07:00
67fb4f6cdb
Merge remote-tracking branch 'nccl/master' into develop
...
[ROCm/rccl commit: 26a0fd2517
2020-06-09 17:40:11 -07:00
89072c82e5
2.7.3-1
...
Add support for A100 GPU and related platforms.
Add support for CUDA 11.
Add support for send/receive operations (beta).
[ROCm/rccl commit: 5949d96f36
2020-06-08 09:31:44 -07:00
3077902de6
Fixing CI as install.sh script should not install dependencies without user request ( #217 )
...
[ROCm/rccl commit: 87db65f22d
2020-06-05 11:04:03 -06:00
8e43b854f1
Making hip-clang the default compiler; documentation update ( #216 )
...
* Making hip-clang the default compiler; documentation update
* Adding back --hip-clang to install.sh as a silent option for CI
[ROCm/rccl commit: dc403e0ca2
2020-06-04 11:58:27 -06:00
55d3b0aa68
Merge pull request #214 from wenkaidu/gdr
...
Use cached value for detecting GDR support only once
[ROCm/rccl commit: 2a4514772c
2020-05-22 13:36:23 -07:00
0108a1219d
Use cached value for detecting GDR support only once
...
[ROCm/rccl commit: 67c8e72ce3
2020-05-22 17:19:10 +00:00
730fddfcf3
Merge pull request #212 from wenkaidu/version
...
Report HIP version in logs
[ROCm/rccl commit: 957be85944
2020-05-20 16:25:54 -07:00
b449a3b36d
Report HIP version in logs
...
[ROCm/rccl commit: e41ab173cf
2020-05-20 18:15:32 +00:00
fbd7d2d0d7
Merge pull request #210 from wenkaidu/unroll
...
Revert "Tuning the inline and unroll to reduce the scratch usage"
[ROCm/rccl commit: af703877cf
2020-05-15 15:35:27 -07:00
e7eff47be4
Revert "Tuning the inline and unroll to reduce the scratch usage"
...
This reverts commit d8a06589c9ca493a6b51
2020-05-15 14:15:40 -07:00
5780637fff
Merge pull request #209 from wenkaidu/hip-clang
...
Rename files which only diffs in extension
[ROCm/rccl commit: c245f1507e
2020-05-15 13:51:12 -07:00
61a82ff572
Merge pull request #208 from wenkaidu/perf_xgmi
...
Give preference to path with more XGMI connections
[ROCm/rccl commit: 706de76046
2020-05-15 10:07:22 -07:00
c2afb1f4ca
Rename files which only diffs in extension
...
[ROCm/rccl commit: e7b36304c8
2020-05-15 09:16:32 -07:00
da05861904
Merge pull request #207 from wenkaidu/hip-clang
...
rccl-prim-test: add flags when calling hipExtLaunchMultiKernelMultiDe…
[ROCm/rccl commit: ca4987e5fb
2020-05-14 18:31:56 -07:00
27519fd019
Give preference to path with more XGMI connections
...
[ROCm/rccl commit: b3c9852634
2020-05-14 15:33:16 -07:00
ced9958094
rccl-prim-test: add flags when calling hipExtLaunchMultiKernelMultiDevice in hip-clang
...
[ROCm/rccl commit: f1058b6353
2020-05-12 23:54:07 +00:00
5b01a901b7
Merge pull request #206 from stanleytsang-amd/develop
...
Updating RCCL documentation
[ROCm/rccl commit: 66a9f11910
2020-05-12 17:24:40 -06:00
abba9d5504
Restoring doxygen documentation to nccl.h.in.
...
[ROCm/rccl commit: 787ac13486
2020-05-12 22:03:31 +00:00
e35e4d3401
Updating README and readthedocs documentation.
...
[ROCm/rccl commit: b59b9d328b
2020-05-12 20:11:49 +00:00
d1cd60fbf4
Merge pull request #205 from wenkaidu/bf16
...
Update rccl_bfloat16.h to match rocBLAS
[ROCm/rccl commit: 52752aba6e
2020-05-11 09:55:06 -07:00
5ba51914f1
Update rccl_bfloat16.h to match rocBLAS
...
[ROCm/rccl commit: d5a07a7b5c
2020-05-08 22:48:07 +00:00
0d642448d7
Merge pull request #204 from wenkaidu/launch_flags
...
Set flags when calling hipExtLaunchMultiKernelMultiDevice in hip-clang
[ROCm/rccl commit: 94d16c0f0a
2020-05-08 11:20:50 -07:00
069322d05a
Set flags when calling hipExtLaunchMultiKernelMultiDevice in hip-clang
...
[ROCm/rccl commit: 24ea2ef6dd
2020-05-08 15:57:14 +00:00
65ad48404b
Merge remote-tracking branch 'upstream/master' into develop
...
[ROCm/rccl commit: 33c23fdcda
2020-04-29 16:12:37 -07:00
75863c97a4
Refactoring packaging ( #193 )
...
[ROCm/rccl commit: 308e96877e
2020-04-29 16:24:21 -06:00
2badecfa20
Merge pull request #199 from wenkaidu/para_jobs
...
Enable parallel jobs for hip-clang build
[ROCm/rccl commit: 914b6ca27c
2020-04-29 13:49:48 -07:00
fa25190ddb
Junit test storage call corrected ( #197 )
...
* Focus testing on Centos for now
* storing junit
* Reducing test suite to Ubuntu
[ROCm/rccl commit: 65390f9872
2020-04-29 14:22:03 -06:00
52096f2cb6
Enable parallel jobs for hip-clang build
...
[ROCm/rccl commit: 3f471ab5b1
2020-04-29 17:58:16 +00:00
c0c0e92ef5
Adding NCCL_DEBUG=INFO Logging to CI ( #196 )
...
[ROCm/rccl commit: 6b1d70b03b
2020-04-27 15:12:15 -06:00
3460376850
Merge pull request #195 from wenkaidu/sync_nccl
...
Sync up with NCCL
[ROCm/rccl commit: f7c27c6c9f
2020-04-27 11:45:05 -07:00
779ee97ada
topo_expl: fix build error
...
[ROCm/rccl commit: 5743c6b7d2
2020-04-27 17:17:05 +00:00
9813d67cd1
Merge remote-tracking branch 'nccl/master' into HEAD
...
[ROCm/rccl commit: c4edc257b0
2020-04-27 17:16:54 +00:00
5f57e6b466
Merge pull request #194 from wenkaidu/search
...
Fix incorrect next device ID in PCI ordered search
[ROCm/rccl commit: cf5070f6c0
2020-04-27 09:54:09 -07:00
7b7f781658
Fix incorrect next device ID in PCI ordered search
...
[ROCm/rccl commit: edb49ed2d5
2020-04-25 01:01:13 +00:00
b9acac2db6
Enabling CI Testing Again ( #192 )
...
Adding CI support based on AMD internal CI refactor.
[ROCm/rccl commit: cc66dd46e9
2020-04-24 10:36:57 -06:00
eebc6f2844
Adding option to re-use streams instead of re-creating per topology
...
[ROCm/rccl commit: 339bf9ff19
2020-04-23 15:53:40 +00:00
c43022b9d8
Fix crash when only a subset of GPUs are visible within a container.
...
Fixes #326 .
[ROCm/rccl commit: f36540f55a
2020-04-17 10:03:14 -07:00
5df2502deb
Improve robustness of PCI detection
...
Fallback to default values when class/speed is unknown.
[ROCm/rccl commit: 23a9fbb788
2020-04-16 14:27:50 -07:00
033f8b7398
Merge pull request #191 from wenkaidu/gfx803
...
Revert "Temporary disable 0x803 target due to build error"
[ROCm/rccl commit: c017f6e900
2020-04-16 09:27:24 -07:00
acd868dfcb
Fix wrong variable name "slice" to "chunk"
...
https://github.com/NVIDIA/nccl/issues/287
[ROCm/rccl commit: a783484ab5
2020-04-14 19:00:51 -07:00
728cf9ee10
Revert "Temporary disable 0x803 target due to build error"
...
This reverts commit 8b1ce44c2a5170bd1c02
2020-04-14 16:58:41 +00:00
Wenkai Du
Sylvain Jeaugey
saadrahim
Stanley Tsang
Wenkai Du
Saad Rahim
Gilbert Lee
aokomoriuta