-
Notifications
You must be signed in to change notification settings - Fork 494
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.
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
add SyncBatchNorm #387
base: master
Are you sure you want to change the base?
add SyncBatchNorm #387
Conversation
- add byteps.torch.SyncBatchNorm - add BYTEPS_SYNC_BN_GLOBAL to choose between global sync and node local sync. defaults to node local sync - add node local allgather Signed-off-by: yulu.jia <[email protected]>
tests/test_bps_torch_syncbn.py
Outdated
skip_or_fail_gpu_test(self, "No GPUs available") | ||
|
||
def test_horovod_sync_batch_norm(self): | ||
print("xxx starting the test") |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Shall we remove this or make it more formal?
tests/test_bps_torch_syncbn.py
Outdated
|
||
bps.init() | ||
torch.cuda.set_device(bps.local_rank()) | ||
print("xxx my local rank is ", bps.rank()) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Same as above
For future reference: https://github.com/facebookresearch/fairscale/pull/662/files |
|
||
Arguments: | ||
tensor: A tensor to allgather. | ||
name: A name of the allgather operation. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
document node_local
?
- add byteps.torch.SyncBatchNorm - add BYTEPS_SYNC_BN_GLOBAL to choose between global sync and node local sync. defaults to node local sync - add node local allgather Signed-off-by: yulu.jia <[email protected]> Signed-off-by: Yulu Jia <[email protected]>
Signed-off-by: yulu.jia [email protected]