George Stykalin a33888dbd1 Use NCCL_DEBUG_FILE for transport logging ...

Replace os.dup2 stderr redirect with NCCL_DEBUG_FILE to avoid
interfering with NCCL's internal logging. NCCL writes TRACE output
to a temp file, then we filter and print only transport-related
lines (NET/IB, GPU Direct, Channel, via NET/Socket).

Co-Authored-By: Claude <noreply@anthropic.com>

2026-04-16 16:38:58 +03:00

..

connection-check-default.py

update graphs

2026-01-26 16:13:01 +03:00

connection-check.py

update dags

2026-01-26 15:49:45 +03:00

test-train-pytorch.py

Use NCCL_DEBUG_FILE for transport logging

2026-04-16 16:38:58 +03:00

test-train-tensorflow.py

init commit

2026-01-12 16:25:19 +03:00

train.py

update dags

2026-04-15 12:10:19 +03:00