Qinghua Zhou a7c364beb8 nccl/rccl integration (#469) ...

Use dlopen to load nccl/rccl Apis from shared library to
enable Allgather, Allreduce, Broadcast, ReduceScatter fallback to nccl/rccl operations.

Add three related environment variables
-x MSCCLPP_ENABLE_NCCL_FALLBACK=TRUE
-x MSCCLPP_NCCL_LIB_PATH=/path/libnccl.so/librccl.so
-x MSCCLPP_FORCE_NCCL_FALLBACK_OPERATION="allreduce,allgather,broadcast,reducescatter" or "all"
By default, if MSCCLPP_FORCE_NCCL_FALLBACK_OPERATION is not specified, all these operations will be fallback to nccl/rccl apis.
---------

Co-authored-by: Binyang Li <binyli@microsoft.com>

2025-03-20 11:31:37 -07:00

..

mscclpp

nccl/rccl integration (#469)

2025-03-20 11:31:37 -07:00

CMakeLists.txt

Add Python get_include() (#141)

2023-07-25 10:23:16 -07:00