Skip to content

Bugfix - Use torchrun --standalone for single-node torch.distributed runs #839

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.

Sign up for GitHub

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

Open
polarG wants to merge 2 commits into
base: main
Choose a base branch
from
Open

Bugfix - Use torchrun --standalone for single-node torch.distributed runs #839

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
c66f82b
Fix single-node torch.distributed: use torchrun --standalone
polarG Jul 26, 2026
3b59d5e
CI/CD - Only cap setuptools below 66 for Python 3.7
Jul 31, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 3 additions & 1 deletion setup.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -19,8 +19,10 @@
print(f'Python {sys.version_info.major}.{sys.version_info.minor} detected.')
if sys.version_info[:2] < (3, 11):
import pkg_resources
# setuptools 66+ no longer supports Python 3.7, newer runtimes work with any recent version.
setuptools_req = 'setuptools>=45, <66' if sys.version_info[:2] < (3, 8) else 'setuptools>=45'
try:
pkg_resources.require(['pip>=18', 'setuptools>=45, <66'])
pkg_resources.require(['pip>=18', setuptools_req])
except (pkg_resources.VersionConflict, pkg_resources.DistributionNotFound):
print(
'\033[93mTry update pip/setuptools versions, for example, '
Expand Down
5 changes: 4 additions & 1 deletion superbench/runner/runner.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -154,8 +154,11 @@ def __get_mode_command(self, benchmark_name, mode, timeout=None):
elif mode.name == 'torch.distributed':
# TODO: replace with torch.distributed.run in v1.9
# TODO: only supports node_num=1 and node_num=all currently
# For single-node runs use torchrun's standalone rendezvous, which binds a random free
# port instead of the fixed default (29500). This avoids transient EADDRINUSE failures
# when consecutive distributed benchmarks reuse the same rendezvous port.
torch_dist_params = (
'' if 'node_num' in mode and mode.node_num == 1 else
'--standalone ' if 'node_num' in mode and mode.node_num == 1 else
'--nnodes=$NNODES --node_rank=$NODE_RANK --master_addr=$MASTER_ADDR --master_port=$MASTER_PORT '
)

Expand Down
1 change: 1 addition & 0 deletions tests/runner/test_runner.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -141,6 +141,7 @@ def test_get_mode_command(self):
'expected_command': (
'torchrun '
'--no_python --nproc_per_node=8 '
'--standalone '
f'sb exec --output-dir {self.sb_output_dir} -c sb.config.yaml -C superbench.enable=foo '
'superbench.benchmarks.foo.parameters.distributed_impl=ddp '
'superbench.benchmarks.foo.parameters.distributed_backend=nccl'
Expand Down
Loading