Skip to content

Fix ddp_spawn -> ddp fallback logic when on LSF cluster #15657

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

Merged
merged 7 commits into from
Nov 12, 2022
Merged

Fix ddp_spawn -> ddp fallback logic when on LSF cluster #15657

Show file tree
Hide file tree
Changes from 3 commits
Commits
Show all changes
7 commits
Select commit Hold shift + click to select a range
f297214
lsf-env-fix
Atharva-Phatak Nov 12, 2022
f80d769
[pre-commit.ci] auto fixes from pre-commit.com hooks
pre-commit-ci[bot] Nov 12, 2022
f431142
fix typo
Atharva-Phatak Nov 12, 2022
973d69e
update tests
awaelchli Nov 12, 2022
35830b0
changelog
awaelchli Nov 12, 2022
1913396
Merge branch 'master' into lsf-env
awaelchli Nov 12, 2022
6c1caea
[pre-commit.ci] auto fixes from pre-commit.com hooks
pre-commit-ci[bot] Nov 12, 2022
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
5 changes: 4 additions & 1 deletion src/pytorch_lightning/trainer/connectors/accelerator_connector.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -614,7 +614,10 @@ def _check_strategy_and_fallback(self) -> None:
strategy_flag = "" if isinstance(self._strategy_flag, Strategy) else self._strategy_flag

if strategy_flag in ("ddp_spawn", "ddp_spawn_find_unused_parameters_false") and (
TorchElasticEnvironment.detect() or KubeflowEnvironment.detect() or SLURMEnvironment.detect()
TorchElasticEnvironment.detect()
or KubeflowEnvironment.detect()
or SLURMEnvironment.detect()
or LSFEnvironment.detect()
):
strategy_flag = "ddp"
if strategy_flag == "dp" and self._accelerator_flag == "cpu":
Expand Down
20 changes: 20 additions & 0 deletions tests/tests_pytorch/trainer/connectors/test_accelerator_connector.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -597,6 +597,26 @@ def test_strategy_choice_ddp_cpu_slurm(cuda_count_0, strategy):
assert trainer.strategy.local_rank == 0


@mock.patch.dict(
os.environ,
{
"LSB_JOBID": "1",
"LSB_DJOB_RANKFILE": "SOME_RANK_FILE",
"JSM_NAMESPACE_LOCAL_RANK": "1",
"JSM_NAMESPACE_SIZE": "20",
"JSM_NAMESPACE_RANK": "1",
},
)
@mock.patch("pytorch_lightning.strategies.DDPStrategy.setup_distributed", autospec=True)
@pytest.mark.parametrize("strategy", ["ddp", DDPStrategy()])
def test_dist_backend_accelerator_mapping(cuda_count_0, strategy):
trainer = Trainer(fast_dev_run=True, strategy=strategy, accelerator="cpu", devices=2)
assert isinstance(trainer.acclerator, CPUAccelerator)
assert isinstance(trainer.strategy, DDPStrategy)
assert isinstance(trainer.strategy.cluster_environment, LSFEnvironment)
assert trainer.strategy.local_rank == 0


@RunIf(min_torch="1.12")
def test_check_native_fsdp_strategy_and_fallback():
with pytest.raises(
Expand Down