Skip to content

don't gather partitioned activations for mp size 1 #2454

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
tjruwase merged 3 commits into from
Nov 4, 2022
Merged

don't gather partitioned activations for mp size 1 #2454

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
95f0a4a
don't gather partitioned activations for mp size 1
guoyejun Oct 29, 2022
982a048
add inline comment for the change
guoyejun Nov 1, 2022
420c3bd
Merge branch 'master' into checkpointing
tjruwase Nov 4, 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
6 changes: 6 additions & 0 deletions deepspeed/runtime/activation_checkpointing/checkpointing.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -270,6 +270,12 @@ def gather_partitioned_activations(tensors, device=None):
inputs.append(item)
continue

# don't need to do all_gather if model parallel size is 1
if mp_size == 1:
item = item.view(list(size.numpy()))
inputs.append(item)
continue

partition_size = item.numel()
tensor_size = partition_size * mp_size
if device is not None:
Expand Down