Skip to content

infra: add single-instance, multi-process Horovod test for local GPU #389

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 1 commit into from
Jun 17, 2020
Merged

infra: add single-instance, multi-process Horovod test for local GPU #389

Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
39 changes: 26 additions & 13 deletions test/integration/local/test_horovod.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -22,25 +22,38 @@
RESOURCE_PATH = os.path.join(os.path.dirname(__file__), '..', '..', 'resources')


@pytest.mark.skip_cpu
@pytest.mark.skip_generic
def test_distributed_training_horovod_gpu(
sagemaker_local_session, image_uri, tmpdir, framework_version
):
_test_distributed_training_horovod(
1, 2, sagemaker_local_session, image_uri, tmpdir, framework_version, 'local_gpu'
)


@pytest.mark.skip_gpu
@pytest.mark.skip_generic
@pytest.mark.parametrize('instances, processes', [
[1, 2],
(2, 1),
(2, 2),
(5, 2)])
def test_distributed_training_horovod_basic(instances,
processes,
sagemaker_local_session,
image_uri,
tmpdir,
framework_version):
@pytest.mark.parametrize(
'instances, processes', [(1, 2), (2, 1), (2, 2), (5, 2)]
)
def test_distributed_training_horovod_cpu(
instances, processes, sagemaker_local_session, image_uri, tmpdir, framework_version
):
_test_distributed_training_horovod(
instances, processes, sagemaker_local_session, image_uri, tmpdir, framework_version, 'local'
)


def _test_distributed_training_horovod(
instances, processes, session, image_uri, tmpdir, framework_version, instance_type
):
output_path = 'file://%s' % tmpdir
estimator = TensorFlow(
entry_point=os.path.join(RESOURCE_PATH, 'hvdbasic', 'train_hvd_basic.py'),
role='SageMakerRole',
train_instance_type='local',
sagemaker_session=sagemaker_local_session,
train_instance_type=instance_type,
sagemaker_session=session,
train_instance_count=instances,
image_name=image_uri,
output_path=output_path,
Expand Down