Use matching MPI datatype for size_t reduction#142

romerojosh merged 2 commits into

NVIDIA:mainNVIDIA/cuDecomp:mainfrom

fallintoplace:fix/nvshmem-size-t-allreducefallintoplace/cuDecomp:fix/nvshmem-size-t-allreduceCopy head branch name to clipboard