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.

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

Jump to bottom

Enable parallel training UT in GitHub CI. #1075

Merged
merged 4 commits into from
Sep 1, 2021
Merged

Enable parallel training UT in GitHub CI. #1075

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
c8cb61c
Remove dupliated setting of `allow_growth` in trainer.
shishaochen Aug 31, 2021
5c8f18b
Make parallel training UT independent of its working folder.
shishaochen Aug 31, 2021
d38338f
Skip parallel-training tests when there is only 1 GPU card.
shishaochen Aug 31, 2021
4280000
Enable parallel training UT in GitHub CI.
shishaochen Aug 31, 2021
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: 5 additions & 0 deletions .github/workflows/test_python.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -71,4 +71,9 @@ jobs:
CXX: g++-${{ matrix.gcc }}
TENSORFLOW_VERSION: ${{ matrix.tf }}
- run: dp --version
- name: Prepare parallel runtime
if: ${{ matrix.tf == '' }}
run: |
sudo apt install libopenmpi-dev openmpi-bin
HOROVOD_WITHOUT_GLOO=1 HOROVOD_WITH_TENSORFLOW=1 pip install horovod mpi4py
- run: pytest --cov=deepmd source/tests && codecov
1 change: 0 additions & 1 deletion deepmd/train/trainer.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -397,7 +397,6 @@ def _init_session(self):
config = get_tf_session_config()
device, idx = self.run_opt.my_device.split(":", 1)
if device == "gpu":
config.gpu_options.allow_growth = True
config.gpu_options.visible_device_list = idx
self.sess = tf.Session(config=config)

Expand Down
7 changes: 5 additions & 2 deletions source/tests/test_parallel_training.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -18,9 +18,12 @@ def setUp(self):
def test_two_workers(self):
command = 'horovodrun -np 2 dp train -m workers ' + self.input_file
penv = os.environ.copy()
if len(get_gpus() or []) > 1:
num_gpus = len(get_gpus() or [])
if num_gpus > 1:
penv['CUDA_VISIBLE_DEVICES'] = '0,1'
popen = sp.Popen(command, shell=True, env=penv, stdout=sp.PIPE, stderr=sp.STDOUT)
elif num_gpus == 1:
raise unittest.SkipTest("At least 2 GPU cards are needed for parallel-training tests.")
popen = sp.Popen(command, shell=True, cwd=str(tests_path), env=penv, stdout=sp.PIPE, stderr=sp.STDOUT)
for line in iter(popen.stdout.readline, b''):
if hasattr(line, 'decode'):
line = line.decode('utf-8')
Expand Down