2.10.3-1

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.

makefiles/common.mk

@@ -55,7 +55,7 @@ CXXFLAGS   := -DCUDA_MAJOR=$(CUDA_MAJOR) -DCUDA_MINOR=$(CUDA_MINOR) -fPIC -fvisi
 # Maxrregcount needs to be set accordingly to NCCL_MAX_NTHREADS (otherwise it will cause kernel launch errors)
 # 512 : 120, 640 : 96, 768 : 80, 1024 : 60
 # We would not have to set this if we used __launch_bounds__, but this only works on kernels, not on functions.
-NVCUFLAGS  := -ccbin $(CXX) $(NVCC_GENCODE) -std=c++11 -Xptxas -maxrregcount=96 -Xfatbin -compress-all
+NVCUFLAGS  := -ccbin $(CXX) $(NVCC_GENCODE) -std=c++11 --expt-extended-lambda -Xptxas -maxrregcount=96 -Xfatbin -compress-all
 # Use addprefix so that we can specify more than one path
 NVLDFLAGS  := -L${CUDA_LIB} -lcudart -lrt
 

makefiles/version.mk

@@ -1,6 +1,6 @@
 ##### version
 NCCL_MAJOR   := 2
-NCCL_MINOR   := 9
-NCCL_PATCH   := 9
+NCCL_MINOR   := 10
+NCCL_PATCH   := 3
 NCCL_SUFFIX  :=
 PKG_REVISION := 1