Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

P2P test fails #90

Open
Aidyn-A opened this issue Jun 16, 2022 · 0 comments
Open

P2P test fails #90

Aidyn-A opened this issue Jun 16, 2022 · 0 comments

Comments

@Aidyn-A
Copy link

Aidyn-A commented Jun 16, 2022

Environment:

ucx: 1.12.1
ucc: 1.0.0
cuda: cuda11.7
gcc: gcc-9.4.0
pytorch: nightly 
open mpi: 4.1.2rc4

The command start_test.sh torch_pt2pt_test.py --backend ucc deadlocks. After waiting a while the following lo shows up

Traceback (most recent call last):
Traceback (most recent call last):
  File "torch_pt2pt_test.py", line 32, in <module>
  File "torch_pt2pt_test.py", line 28, in <module>
    dist.recv(tensor_test, src=0, tag=0, group=pg)
  File "/opt/conda/lib/python3.8/site-packages/torch/distributed/distributed_c10d.py", line 1007, in recv
    dist.send(tensor_test, dst=dst + 1, tag=0, group=pg)
  File "/opt/conda/lib/python3.8/site-packages/torch/distributed/distributed_c10d.py", line 964, in send
    pg.recv([tensor], group_src_rank, tag).wait()
    group.send([tensor], group_dst_rank, tag).wait()
RuntimeError: Socket Timeout
RuntimeError: Socket Timeout
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

1 participant