viable/strict/1787959238: Wire symmetric memory operations to the distributed watchdog (#193834)
- PyTorch 近 90 天出现 190 次
- PyTorch 近 90 天第 187 次发版
- 上一次：同一天稍早 · viable/strict/1787957414
发生了什么
Summary Integrates the torch.distributed._watchdog timeout machinery into symmetric memory's blocking and stream-based operations so that hangs in multi-process rendezvous or one-sided NCCL ops are detected and surfaced instead of blocking silently. Test Plan python test/distributed/test_symmetric_memory.py -k SymmMemWatchdogUnitTest torchrun --nproc_per_node=2 test/distributed/test_nccl.py -k NCCLSymmMemWatchdogTes…
摘要按规则整理自下方来源原文