Add support for bfloat16.
Add ncclAvg reduction operation.
Improve performance for aggregated operations.
Improve performance for tree.
Improve network error reporting.
Add NCCL_NET parameter to force a specific network.
Add NCCL_IB_QPS_PER_CONNECTION parameter to split IB traffic onto multiple queue pairs.
Fix topology detection error in WSL2.
Fix proxy memory elements affinity (improve alltoall performance).
Fix graph search on cubemesh topologies.
Fix hang in cubemesh during NVB connections.
[ROCm/rccl commit: 7e51592129]
* Select sendrecv path based on collective data size
* Add comments on packing and unpacking group field
* Toggling RCCL_P2P_NET_DISABLE in combined calls unit tests
[ROCm/rccl commit: 6dcae8a459]
* Re-enabling mp unit tests
* Fixing shared memory leak and other bugs related to shared mem for MP unit tests
* Revert 43bfbfc97bf9edbae1f386d461439091618ff8ed
* Further tightening up unlinks
* Moving test check macros to separate header file
* Tightening up shared memory unlinking for clique kernels, add munmap for host barrier for MP unit tests
* Updating new MP unit test
* Fixing mqueue bug
* Fixing memory leak in MP unit tests
[ROCm/rccl commit: 0b2bfdd6d8]
Fix crash when setting NCCL_MAX_P2P_NCHANNELS below nchannels.
Fix hang during sendrecv dynamic NVB connection establishment on
cubemesh topologies.
Add environment variable to only use SHARP on communicators beyond
a given number of ranks.
Add debug subsystem to trace memory allocations.
Fix compilation with TRACE=1. (Issue #505)
[ROCm/rccl commit: 3fec2fa5ee]
* 2.9.6-1
Add support for CUDA graphs.
Fuse BCM Gen4 switches to avoid suboptimal performance on some platforms. Issue #439.
Fix bootstrap issue caused by connection reordering.
Fix CPU locking block.
Improve CollNet algorithm.
Improve performance on DGX A100 for communicators with only one GPU per node.
* Clique tuning upgrade (#352) (#19)
* Enabling clique for any XGMI-connected topology, adding tuning
* Updating CHANGELOG for clique tuning
* Re-working clique barrier system to work on multi-process / multi-gpu
Co-authored-by: Sylvain Jeaugey <sjeaugey@nvidia.com>
Co-authored-by: gilbertlee-amd <44450918+gilbertlee-amd@users.noreply.github.com>
[ROCm/rccl commit: 6021329af0]
* Enabling clique for any XGMI-connected topology, adding tuning
* Updating CHANGELOG for clique tuning
* Re-working clique barrier system to work on multi-process / multi-gpu
[ROCm/rccl commit: e796b1645c]
Fix memory leaks.
Fix crash in bootstrap error case.
Fix Collnet clean-up issue.
Make PCI switch vendor/device optional for XML injection.
Add support for nvidia-peermem module.
[ROCm/rccl commit: ca8485b0d0]
* Enabling clique for any XGMI-connected topology, adding tuning
* Updating CHANGELOG for clique tuning
* Re-working clique barrier system to work on multi-process / multi-gpu
[ROCm/rccl commit: 9d7232c091]
* Fixing message queue leak.
* Using POSIX implementation of Message Queues
* Adding unlink to msgqueue
* MsgQueue update
* Adding timeout check to msgqueue broadcast; tightening up system checks
* Removing unnecessary code
* Removing extra argument from print
* Adding explicit msg queue close call to all other ranks
[ROCm/rccl commit: 70597789d0]
Ke Wen
Wenkai Du
Eiden Yoshida
gilbertlee-amd
Stanley Tsang
Wenkai Du
Sylvain Jeaugey
saadrahim