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

Fix unnecessary sync_memops cuda pointer attr in GDR #12170

Merged
merged 1 commit into from Aug 10, 2017
Merged

Fix unnecessary sync_memops cuda pointer attr in GDR #12170

merged 1 commit into from Aug 10, 2017

Conversation

byronyi
Copy link
Contributor

@byronyi byronyi commented Aug 10, 2017

See the comments here. I've tested on my local boxes and it seems removing CU_POINTER_ATTRIBUTE_SYNC_MEMOPS slightly improves the training throughput while introducing no data race.

@tensorflow-jenkins
Copy link
Collaborator

Can one of the admins verify this patch?

@rmlarsen rmlarsen requested a review from zheng-xq August 10, 2017 15:41
@rmlarsen rmlarsen requested review from poxvoculi and removed request for zheng-xq August 10, 2017 15:42
@rmlarsen rmlarsen assigned poxvoculi and unassigned zheng-xq Aug 10, 2017
@rmlarsen
Copy link
Member

@tensorflow-jenkins test this please

@rmlarsen
Copy link
Member

Test failure is unrelated.

@rmlarsen rmlarsen merged commit 703fd44 into tensorflow:master Aug 10, 2017
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@poxvoculi poxvoculi poxvoculi approved these changes

Assignees

@poxvoculi poxvoculi

Labels
cla: yes
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.

None yet
6 participants
@byronyi @tensorflow-jenkins @rmlarsen @poxvoculi @googlebot @zheng-xq