Thanks to visit codestin.com
Credit goes to github.com

Skip to content

fix #7188 #7371

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
loadams merged 6 commits into from
Aug 4, 2025
Merged

fix #7188 #7371

Changes from all commits
Commits
Show all changes
6 commits
Select commit Hold shift + click to select a range
2e99dcb
fix data race when swap ipg_index in zero 2
lpnpcs Jul 30, 2025
9f256ce
Merge branch 'master' into fix_grad_nan
lpnpcs Jul 31, 2025
bd33c62
Merge branch 'master' into fix_grad_nan
loadams Jul 31, 2025
c13318f
Merge branch 'master' into fix_grad_nan
lpnpcs Aug 1, 2025
91ab4d3
Merge branch 'master' into fix_grad_nan
sfc-gh-truwase Aug 2, 2025
290f6c7
Merge branch 'master' into fix_grad_nan
hwchen2017 Aug 4, 2025
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
3 changes: 3 additions & 0 deletions deepspeed/runtime/zero/stage_1_and_2.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -995,6 +995,9 @@ def reduce_independent_p_g_buckets_and_remove_grads(self, param, i):
self.report_ipg_memory_usage("In ipg_remove_grads before reduce_ipg_grads", param.numel(), param.dtype)
self.reduce_ipg_grads()
if self.contiguous_gradients and self.overlap_comm:
if not get_accelerator().resolves_data_dependency():
self.reduction_stream.wait_stream(get_accelerator().current_stream())
Comment thread
tjruwase marked this conversation as resolved.
get_accelerator().current_stream().wait_stream(self.reduction_stream)
# Swap index between 0 and 1
bucket.index = 1 - bucket.index
self.report_ipg_memory_usage("In ipg_remove_grads after reduce_ipg_grads", param.numel(), param.dtype)
Expand Down