81ddf9de89
Merge remote-tracking branch 'nccl/v2.19' into develop
2024-01-24 15:25:33 -08:00
7e25d5bc55
Use new HIP graph API compatible with CUDA 11030 ( #991 )
...
* Use new HIP graph API compatible with CUDA 11030
* Update dependency to ROCm 6.1
* Fix single stream use case
2024-01-21 19:00:50 -08:00
70bbeb4773
Fix MSCCL multi-node ( #1032 )
...
1) Move needsProxy initialization before mscclSetupConnections since the latter
will revise it later.
2) Remove mscclAvailable check in net.cc since it's no more required and caused
non-shared buffer allocated for MSCCL which is not expected.
2024-01-05 17:03:43 -08:00
0a53077c9c
Improve MSCCL algorithms ( #1023 )
2024-01-03 14:51:34 -08:00
f4858e14b2
rearranged how the min and max functions are part of msccl ( #1025 )
...
* rearranged how the min and max functions are part of msccl
* added more coverage on in place graph tests
2023-12-21 08:58:33 -07:00
655742a3a6
Fully disable MSCCL when machine is not matched ( #1017 )
...
* Disable MSCCL algorithm meta loading when machine is not matched
* fully disable init
* fix potential segfault
2023-12-13 08:36:21 -08:00
c002f20029
Fix MSCCL scratch allocation ( #1010 )
2023-12-08 17:47:10 -06:00
12c08fc52a
msccl: build same number of kernels as in ROCm 5.7 ( #1005 )
...
Removed fullOps kernels from build
2023-12-07 13:36:04 -06:00
293f0fb752
Use a map to host scratch buffers ( #1004 )
...
* Use a map to host scratch buffers
* Address review feedbacks. Deliberately keep mscclSetupScratch function.
2023-12-05 13:15:28 -06:00
e44e112a17
Fix mscclAlgoHandle not initialized issue ( #995 )
2023-12-01 07:58:01 -08:00
4bb0b4a380
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
2023-11-30 09:47:20 -08:00
56ce9ef05f
recreated pr 914 to work with current develop branch ( #979 )
2023-11-28 16:33:47 -07:00
569d3f7d59
msccl: allocate scratch as ext-scope fine-grained ( #968 )
2023-11-16 09:57:25 -06:00
7fc891bc8d
Fix MSCCL work FIFO allocation with HIP graph enabled ( #967 )
2023-11-15 16:43:28 -08:00
198f14923b
Check to support older ROCm versions ( #963 )
2023-11-15 12:36:31 -07:00
7ae95db5b8
Optimize MSCCL all-gather algorithms for gfx942 ( #964 )
2023-11-15 08:18:59 -08:00
5a800e00cd
msccl: enable basic collective trace ( #959 )
...
To avoid increasing number of kernels, colltrace is only enabled with
RCCL_MSCCL_FORCE_FULLOPS=1
2023-11-08 20:14:28 -08:00
bb84345943
msccl: use 32-bit LDS access and add RCCL_MSCCL_FORCE_FULLOPS ( #953 )
2023-11-03 10:38:02 -07:00
988efe605a
MSCCL stream fix ( #948 )
2023-11-03 09:10:52 -06:00
f484ff17b9
msccl: add templated kernel ( #945 )
...
* msccl: add templated kernel
* Use defines to improve code readability
* Fix kernel indexing and review feedback
2023-11-02 17:21:53 -07:00
1e5ca6820b
Fix gcnArchName bug in topology dump ( #937 )
...
Signed-off-by: nileshnegi <Nilesh.Negi@amd.com >
2023-10-28 12:30:36 -05:00
4c117e5335
Fix MSCCL work FIFO out-of-bound issue ( #935 )
2023-10-27 11:24:52 -07:00
f22df90e5c
remove gcnArch support ( #920 )
...
Signed-off-by: nileshnegi <Nilesh.Negi@amd.com >
2023-10-26 12:09:15 -05:00
fb0eccb57b
msccl: reduce debug output when using NCCL_DEBUG=INFO ( #932 )
2023-10-25 08:05:19 -07:00
1b80d041cb
Fix incorrect arch name parsing ( #916 )
2023-10-13 10:01:11 -07:00
6d0b5c1e89
Port init_once fix from NCCL ( #915 )
2023-10-13 08:01:12 -07:00
7ee5c1c28b
Change MSCCL kernel signature to allow kernel arguments be preloaded via SGPR ( #911 )
...
* Adding a script that will download/compile/run TransferBench/RCCL/UCX/RCCL-tests/RCCL-Unittests/hip-mpi-testsuite (#895 )
Co-authored-by: Pedram Alizadeh <pmohamma@banff-pla-r27-05.pla.dcgpu >
* Only build gfx941
* demo
* fine tune malloc
* Fix merge errors
* Fix merge errors
* Disable parallel build
* Adopt --amdgpu-kernarg-preload-count
* Revert "Adding a script that will download/compile/run TransferBench/RCCL/UCX/RCCL-tests/RCCL-Unittests/hip-mpi-testsuite (#895 )"
This reverts commit f5e252dddf02a41b4d1bc512f306f45f97166304.
* Revert CMake changes.
* NPKIT changes.
* Remove some license declarations.
* Address code review feedbacks on msccl_kernel_impl.h
* Update CMakeLists.txt
* Add CMake logic to check the existence of --amdgpu-kernarg-preload-count
* Fix NPKIT trace logic.
---------
Co-authored-by: Pedram Alizadeh <pmohamma@amd.com >
Co-authored-by: Pedram Alizadeh <pmohamma@banff-pla-r27-05.pla.dcgpu >
Co-authored-by: Ziyue Yang <ziyyang@microsoft.com >
2023-10-12 20:17:08 -05:00
28d7fe5629
Dma buf support optin ( #905 )
...
* dmaBufSupport Optin added on every part of the code that should invoke it
2023-10-03 03:17:48 -06:00
f9c3dc251e
2.19.1-1
...
Add local user buffer registration for NVLink SHARP.
Add tuning plugin support.
Increase net API to v7 to allow for device-side packet reordering;
remove support for v4 plugins.
Add support for RoCE ECE.
Add support for C2C links.
Better detect SHM allocation failures to avoid crash with Bus Error.
Fix missing thread unlocks in bootstrap (Fixes #936 ).
Disable network flush by default on H100.
Move device code from src/collectives/device to src/device.
2023-09-26 05:50:33 -07:00
6a4d5ec089
Fix crash when NPKit is enabled ( #887 )
2023-09-13 11:00:12 -07:00
e58ec78d35
Gcn arch name ( #886 )
...
We use CMake to determine if we're compiling against a version of ROCm that supports gcnArchName and handles architecture checking appropriately. It includes a few helper functions as drop ins for the functionality we used gcnArch for before; sometimes to enable flags, and sometimes to set frequencies.
2023-09-12 15:34:40 -04:00
e1dc4d5e42
enable hip graph on multi-node ( #884 )
...
* initial checkin
* enable msccl when hip graph is on
* remove the commented out code of msccl enable check
* clean up the code
* remove the msccl HighestTransportType check logic
2023-09-11 15:30:04 -07:00
6230b5f6b3
RCCL should use hipPointerAttribute_t.type ( #872 )
2023-09-05 09:44:12 -06:00
4297315de7
More robust msccl shared directory location discovery ( #868 )
2023-08-30 08:10:14 -06:00
5bcd3768cc
Minor fix for some msccl installations ( #862 )
2023-08-23 13:48:58 -06:00
64c32d1c5b
Disable MSCCL kernels at compile time ( #834 )
...
* Disable MSCCL kernels at compile time
2023-08-02 09:45:18 -06:00
f7dc7b7e6a
Fix MSCCL proxy number of chunks calculation ( #821 )
...
Current number of transmissions parsed from MSCCL algorithm is 1-based value,
but when calculating proxy number of chunks, it's taken as 0-based value.
This commit fixes this issue.
2023-07-26 13:24:49 -07:00
a7fcd58a97
Enable gfx94x ( #808 ) ( #816 )
...
(cherry picked from commit 94da229a7788d74685d1591a4e75a8341de64f41)
2023-07-21 07:31:27 -07:00
848e60b00c
Fix path finding in msccl internal scheduler ( #794 )
2023-07-12 13:31:28 -07:00
ce6a2ffac8
Merge pull request #782 from ROCmSoftwarePlatform/2.18.3
...
Sync up with NCCL 2.18.3
2023-06-29 15:04:16 -07:00
abd0615351
Merge remote-tracking branch 'nccl/master' into develop
2023-06-26 22:51:56 +00:00
bd14ac8b59
ASAN build excluding additional files, Algodir support for share folder
...
* ASAN build excluding additional files, Algodir support for share folder (#786 )
* Algodir support for share folder
2023-06-23 10:57:20 -04:00
9bdf6797a5
fixed npkit size to never be a negative number ( #779 )
2023-06-21 08:26:40 -06:00
ea38312273
2.18.3-1
...
Fix data corruption with Tree/LL128 on systems with 1GPU:1NIC.
Fix hang with Collnet on bfloat16 on systems with less than one NIC
per GPU.
Fix long initialization time.
Fix data corruption with Collnet when mixing multi-process and
multi-GPU per process.
Fix crash when shared memory creation fails.
Fix Avg operation with Collnet/Chain.
Fix performance of alltoall at scale with more than one NIC per GPU.
Fix performance for DGX H800.
Fix race condition in connection progress causing a crash.
Fix network flush with Collnet.
Fix performance of aggregated allGather/reduceScatter operations.
Fix PXN operation when CUDA_VISIBLE_DEVICES is set.
Fix NVTX3 compilation issues on Debian 10.
2023-06-14 01:29:17 -07:00
d52b6c0d24
add DMA_BUF support ( #763 )
...
* add DMA_BUF support
* remove unused libraries in src/init.cc
* change NCCL_ALL to NCCL_INIT
* remove extra pointer functions in transport/net.cc
2023-06-01 12:46:42 -06:00
5a38ff192b
Rework barrier and event code ( #761 )
...
* Rework barrier and event code
* Switch to inline asm
2023-05-31 13:36:51 -07:00
7d6e7bcd7d
revert npkit ( #748 )
2023-05-24 07:41:05 -07:00
11676267b5
fix min, max and avg ( #745 )
2023-05-18 11:02:59 -07:00
ca4a1dfd67
Address review feedbacks and make the flag be disabled by default.
2023-05-17 17:50:25 +00:00
12dba425de
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.
2023-05-17 16:49:42 +00:00
BertanDogancay
Wenkai Du
Ziyue Yang
akolliasAMD
Wen-Heng (Jack) Chung
Bertan Dogancay
Nilesh M Negi
Sylvain Jeaugey
Audrey MP
Andy li
gilbertlee-amd
Wenkai Du
arvindcheru
Wen-Heng (Jack) Chung