Skip to content

Fix distributed launch.py #17571

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
chengduoZH merged 2 commits into from
May 23, 2019
Merged

Fix distributed launch.py #17571

Changes from 1 commit
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
19d04bd
fix distributed launch.py
May 22, 2019
aa4e624
Merge branch 'develop' of https://github.com/PaddlePaddle/Paddle into…
May 23, 2019
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: 2 additions & 2 deletions python/paddle/distributed/launch.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -75,11 +75,11 @@ def start_procs(gpus, entrypoint, entrypoint_args, log_dir):
nranks = num_nodes * gpus
# ======== for dist training =======
gpu_ids = get_gpu_ids(gpus)
for i in gpu_ids:
for i in range(gpus):
curr_env = {}
curr_env.update(default_envs)
curr_env.update({
"FLAGS_selected_gpus": "%d" % i,
"FLAGS_selected_gpus": "%d" % gpu_ids[i],
"PADDLE_TRAINER_ID": "%d" % (node_trainer_id * gpus + i),
"PADDLE_CURRENT_ENDPOINT": "%s:617%d" % (current_ip, i),
# nranks
Expand Down