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]
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]
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]
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.
[ROCm/rccl commit: a46ea10583]
Fix hang in corner cases of alltoallv using point to point send/recv.
Harmonize error messages.
Fix missing NVTX section in the license.
Update README.
[ROCm/rccl commit: 911d61f214]
Optimization for Tree allreduce on A100.
Improve aggregation performance.
Use shared buffers for inter-node send/recv.
Add NVTX profiling hooks.
Accelerate alltoall connections by merging communication for all
channels.
Add support for one hop communication through NVLink, for faster
send/recv communication on cubemesh topologies like DGX-1.
Improve alltoall scheduling to better balance intra/inter node
communication.
Increase send/recv parallelism by 8x, each warp sending or
receiving to a different peer.
Net: move to v4.
Net: make flush operation asynchronous to accelerate alltoall.
Net: define maximum number of requests.
Fix hang when using LL128 protocol after 2^31 steps.
Fix#379 : topology injection failing when using less GPUs than
described in the XML.
Fix#394 : protocol mismatch causing hangs or crashes when using
one GPU per node.
[ROCm/rccl commit: 920dbe5b35]
If a PCI node is the parent of a NIC, treat it as such, regardless of
the PCI class code for the device. This allows non-traditional devices
to act as NICs via the net plugin mechanism.
For consistency, treat GPUs similarly.
[ROCm/rccl commit: 700c0e0f24]
Add $(INCTARGETS) to build dependencies of %.o and $(DEVICELIB).
As there were no dep files during the first build, Make may kick off source
compilation before nccl.h got generated, which leads to occasional build
failures on systems with high core count. The build failure could be
reproduced reliably with a `sleep 5` in $(INCDIR)/nccl.h rule.
[ROCm/rccl commit: 2d8601701d]
Add support for A100 GPU and related platforms.
Add support for CUDA 11.
Add support for send/receive operations (beta).
[ROCm/rccl commit: 5949d96f36]
The reduction tree (tree up) was inverting the NICs to use,
causing performance issue in cases where we are using different
NICs on a given channel.
[ROCm/rccl commit: b5b6c6acdd]
Add support for network collectives.
Add support for XML topology dump/injection.
Add text values for GDR and P2P Levels, including "NVL".
Add speed detection for PCI, Infiniband and Ethernet cards.
Add CPU detection for ARM and AMD CPUs.
Add support for adaptive routing on Infiniband.
Change NET plugin API to v3 : merge PCI path and GPU pointer
capability into a single structure and add other properties.
[ROCm/rccl commit: b221128eca]
Current NCCL code does not abort for failed Flush operations by
underlying network. This may compromise data integrity.
Signed-off-by: Rashika Kheria <rashika@amazon.com>
[ROCm/rccl commit: 6c61492eba]
Clang doesn't understand `optimize("O0")`. It has `noopt`, which GCC doesn't understand. Wrap the difference in a macro.
[ROCm/rccl commit: 8c564e9b57]
Add LL128 Protocol.
Rewrite the topology detection and tree/ring creation (#179). Improve
tree performance by sending/receiving from different GPUs. Add
model-based tuning to switch between the different algorithms and
protocols.
Rework P2P/SHM detection in containers (#155, #248).
Detect duplicated devices and return an error (#231).
Add tuning for GCP
[ROCm/rccl commit: 299c554dcc]
Fixes: #244
SO_RESUEPORT was introduced in Linux 3.9 and later.
This change allows NCCL to compile against older releases.
The functionality is only required if the user is specifying
a NCCL bootstrap address via an environment variable.
[ROCm/rccl commit: 7f2b337e70]
Perform exact matching when the prefix "=" is specified in the NCCL_IB_HCA variable to exclude HCAs mlx5_X[0-9]+ when mlx5_X is specified.
[ROCm/rccl commit: 0b192d2299]
The affinityStr string was not null-terminated but was passed to strlen(3).
Signed-off-by: Felix Abecassis <fabecassis@nvidia.com>
[ROCm/rccl commit: 37e4f8729e]
Makes a change to allow the user to select between the static CUDA
runtime library (default) and the dynamic CUDA runtime library. Does
this by allowing `CUDARTLIB` to be overridden.
[ROCm/rccl commit: 60a586ded9]
John Bachan
Ke Wen
Sylvain Jeaugey
Jonas Zhou
xietingwew
Jack Snyder
Eric Badger
David Addison
Riatre Foo
aokomoriuta
Rashika Kheria
Luke Yeager
Christian Sigg
Cao Zongyan
Hirochika Asai
Felix Abecassis
Rajat Chopra
jakirkham
Gustavo Alvarez