Out-Of-Box Native GPU Allreduce
This feature enables UCC library to detect the NVIDIA NVLink topology and select the best GPU-based algorithms for supported collectives (Allgather/v, Reducescatter/v).
To view the NVLink topology, run nvidia-smi topo -m
To activate this feature, make sure to enable the hierarchical component in UCC using the UCC_CLS environment variable as follows:
-x UCC_CLS=basic,hier.
To view all available UCC items and options, run ucc_info -f