remove the MultiRank examples, the features was never released (because
it didn't work reliably), and it might just cause confusion if somebody
sees it. In additional, the locdation in tools was suboptimal.
[ROCm/rccl commit: 8953a26bcd]
* Restore number of parallel linking jobs
* Dynamically adjust number of linker jobs with limit of 16 jobs max
* Fix typo
* Add cgroup v1 support
[ROCm/rccl commit: ca5341d419]
* Adding performance collection feature in rccl_replayer, and updating MSCCL logging and replayer parsing
* Performance collection feature in rccl_replayer, and updating MSCCL logging and replayer parsing
[ROCm/rccl commit: cf311b71ee]
* adding rocprof parser script
* adding the support for multiple json files
* adding pytorch profiler script
* remove filtering from pytorch log
* adding the addressing the comments and add the feature to parse all kernels
* completing the report for torch profiler
---------
Co-authored-by: Marzieh Berenjkoub <mberenjk@amd.com>
[ROCm/rccl commit: 519843d2cf]
* Template unroll for RCCL kernels
* Adding unroll template arg during CMake hipification
* Reduce linking parallel jobs to avoid OOM in CI
* Workaround issues with UT tests
SWDEV-469533: register spill fix is needed for mainline build
LWPCOMMLIBS-369: cannot enable 112 channels with 80 CUs
Use -parallel-jobs=8 for linking
* CI: do not use -j 16 when building
* CI: use -j 8 when building
* Only reduce parallel linking job for CI extended
* Restore original jenkins command. Change parallel linking jobs in cmake
* Disable MSCCLPP
---------
Co-authored-by: gilbertlee-amd <gilbert.lee@amd.com>
[ROCm/rccl commit: 89349f2ce4]
* Revert "[GRAPH] Use channel shuffling only for IB systems (#1228)"
This reverts commit 8a7dd0e590.
* Revert "Revert "Changing channel stride for MI300X multinode (#1196)" (#1224)"
This reverts commit bb6fab3d8e.
[ROCm/rccl commit: a1ef217b32]
* Only enable MSCCL++ AllReduce for message sizes that are multiples of 32. MSCCL++ does not handle these other sizes.
* Sanitized MSCCL++ logging.
[ROCm/rccl commit: 9cbb3da224]
MSCCL can now run in a multi-threaded configuration. To test in the unit tests, added the ENABLE_OPENMP compile definition flag and the --openmp-test-enable flag to the unit test build script. To activate, set the environment variables UT_MULTITHREADED=1 and UT_PROCESS_MASK=1. Set Jenkins to use this mode.
[ROCm/rccl commit: 0c36d571ea]
* [GRAPH] Use channel shuffling only for IB systems
Signed-off-by: nileshnegi <Nilesh.Negi@amd.com>
* [GRAPH] Define channels=48 for gfx94 RoCE systems
Signed-off-by: nileshnegi <Nilesh.Negi@amd.com>
* [GRAPH] Increase channels for RoCE gfx94 systems
Signed-off-by: nileshnegi <Nilesh.Negi@amd.com>
---------
Signed-off-by: nileshnegi <Nilesh.Negi@amd.com>
[ROCm/rccl commit: 5be3b713ef]
* Fix DMABUF support
* Reduce log output by moving dmabuf allocation details to TRACE
* Enable peer memory GDR support if ib_umem_get_peer is in kernel
[ROCm/rccl commit: 5d7078e383]
Rework core for NVIDIA Trusted Computing
* Compress work structs so that they are shared between channels
* Utilize the full amount of kernel argument space permitted (4k)
before resorting to work fifo.
* Rework the task preprocessing phase.
* Use a separate abortDevFlag which is kept in sync with abortFlag
using cudaMemcpy operations.
* Rename src/include/align.h to src/include/bitops.h
Add lazy connection establishment for collective operations
* Move buffer allocation and connection establishment to the first
collective operation using that algorithm.
* Accelerate init time and reduce memory usage.
* Avoid allocating NVLS buffers if all calls are registered.
* Compute algo/proto in ncclLaunchCollTasksInfo early on.
* Connect peers in ncclCollPreconnectFunc if not connected already.
* Also move shared buffer creation to the first send/recv call.
Accelerate intra-node NVLink detection
* Make each rank only detect NVLinks attached to its GPU.
* Fuse XMLs to reconstruct the full NVLink topology
Add init profiling to report time spend in different init phases.
* Report timings of bootstrap, allgather, search, connect, etc.
* Add new "PROFILE" category for NCCL_DEBUG_SUBSYS.
Add support for PCI p2p on split PCI switches
* Detect split PCI switches through a kernel module exposing
switch information.
* Update the topology XML and graph to add those inter-switch
connections.
Add cost estimation API
* Add a new ncclGroupEndSimulate primitive to return the estimated
time a group would take.
Net/IB: Add separate traffic class for fifo messages
* Add NCCL_IB_FIFO_TC to control the traffic class of fifo messages
independently from NCCL_IB_TC.
Merges PR #1194
Net/IB: Add support for IB router
* Use flid instead of lid if subnets do not match
* Warn if flid is 0
Optimizations and fixes for device network offload (unpack)
* Double the default number of channels
* Cache netDeviceType
* Fix save/increment head logic to enable Tree support.
Support ncclGroupStart/End for ncclCommAbort/Destroy
* Allow Abort/Destroy to be called within a group when managing
multiple GPUs with a single process.
Improve Tuner API
* Provide to the plugin the original cost table so that the plugin
can leave unknown or disabled algo/proto combinations untouched.
* Remove nvlsSupport and collnetSupport.
Do not print version to stdout when using a debug file
* Also print version from all processes with INFO debug level.
Fixes issue #1271
Fix clang warnings in NVTX headers
* Update NVTX headers to the latest version
Fixes issue #1270
Disable port fusion in heterogeneous systems
* Do not fuse ports if a mix of multi-port and single port are detected.
Fix NVLS graphs search for dual NICs.
* Fix NVLS graph search when we have more than one NIC per GPU.
Fix crash with collnetDirect
* Add separate graph search for collnetDirect, testing alltoall paths
and working similarly to the NVLS search.
Fix hang when nodes have different CPU types
* Add the CPU type to the rank peer info.
* Align all ranks on the CPU type after the first allgather.
* Only use the aligned CPU type for all tuning operations.
Fixes issue #1136
Fixes issue #1184
Fix performance of registered send/recv operations
* Allow for single full size operations
* Add INFO to confirm the registration of send/recv buffers.
Move all sync ops to finalize stage
* Ensure ncclCommDestroy is non-blocking if ncclCommFinalize has
been called.
Improve error reporting during SHM segment creation
Improve support of various compilers
Merges PR #1177
Merges PR #1228
Allow net and tuner plugins to be statically linked
* Search for ncclNet or ncclTuner symbols in the main binary.
Merges PR #979
Plugin examples includes cleanup
* Harmonize err.h and common.h usage.
* Add mixed plugin with both net and tuner.
[ROCm/rccl commit: 178b6b7590]
Edgar Gabriel
akolliasAMD
dependabot[bot]
Pedram Alizadeh
Tim
Nilesh M Negi
Ziyue Yang
Wenkai Du
Richard Barnes
Sam Wu
Benjamin Kitor
Joseph Macaranas
corey-derochie-amd
saurabhAMD
mberenjk
Nusrat Islam
amd-jmacaran
Rahul Vaidya
Jack Taylor
Sylvain Jeaugey