【解决】RuntimeError: Distributed package doesn‘t have NCCL built in

现象

python在windows环境下dist.init_process_group(backend, rank, world_size)处报错‘RuntimeError: Distributed package doesn’t have NCCL built in’

原因分析：

windows不支持NCCL backend

解决方法1

import sys

if sys.platform == "win32":
     os.environ["PL_TORCH_DISTRIBUTED_BACKEND"] = "gloo"

解决方法2

import os

os.environ["PL_TORCH_DISTRIBUTED_BACKEND"] = "gloo"