操作系统版本：例如ubuntu20.04 DTK版本：例如DTK24.04.1 驱动版本：例如rock-5.7.1-6.2.18 或者提供如下指令结果rpm -qa | grep rock DCU类型: Z100L 执行程序: import torch import torch.distributed as dist import time import sys def setup(rank, world_size): dist.init_process_group(backend='nccl', rank=rank, world_size=world_size, init_method=f'tcp://127.0.0.1:26001') torch.cuda.set_device(rank) torch.distributed.barrier() def main(rank, world_size): setup(rank, world_size) size = 2048 * 7168 // world_size tensor = torch.randn(size,).cuda() tensor_list = [torch.randn(size,).cuda() for _ in range(world_size)] ``` dist.all_gather(tensor_list, tensor) ``` if __name__ == "__main__": world_size = 2 # ▒~A~G设使▒~T▒两个 GPU torch.multiprocessing.spawn(main, args=(world_size,), nprocs=world_size, join=True) 遇到问题:  程序卡住不动  amd cpu上可以正常运行; intel cpu 卡住不动; 都是一个docker 镜像 py310