Skip to content

/ pytorch

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

Jump to bottom

Fuse operators inside of the (CPU) glu operator. #26418

Closed
VitalyFedyunin opened this issue Sep 18, 2019 · 4 comments
Closed

Fuse operators inside of the (CPU) glu operator. #26418

VitalyFedyunin opened this issue Sep 18, 2019 · 4 comments
Assignees
@gchanan
Labels
good first issue module: cpu topic: bootcamp triaged

Comments

@VitalyFedyunin
Copy link
Contributor

@VitalyFedyunin VitalyFedyunin commented Sep 18, 2019

Current implementation does sequential sigmoid_out and mul_. We can get better performance by fusing this operations together.
@VitalyFedyunin VitalyFedyunin mentioned this issue Sep 18, 2019
Closed
@f2013519

This comment has been minimized.

Sign in to view
Copy link

@f2013519 f2013519 commented Sep 20, 2019

I'll take this up. Is it mostly about adding a at::parallel_for?
@VitalyFedyunin

This comment has been minimized.

Sign in to view
Copy link
Contributor Author

@VitalyFedyunin VitalyFedyunin commented Sep 20, 2019

This can be done after #26410 landed, and will require writing new ternary loop which does sigmoid and mul in one loop.
@f2013519

This comment has been minimized.

Sign in to view
Copy link

@f2013519 f2013519 commented Sep 20, 2019

Sure, I'll wait for that. I've used pytorch before, but I'm new to the codebase.

Can you pls. sketch out roughly how to go about adding this support, like which API's to use, which files may need to updated, or a PR that does something similar?

Thanks!
@gchanan gchanan self-assigned this Sep 22, 2019
@gchanan

This comment has been minimized.

Sign in to view
Copy link
Contributor

@gchanan gchanan commented Sep 23, 2019

I'm actually going to take this because I want an end-to-end change to benchmark.
@gchanan gchanan closed this Feb 27, 2020
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees

@gchanan gchanan

Labels
good first issue module: cpu topic: bootcamp triaged
Projects
None yet
Milestone
No milestone
Linked pull requests

Successfully merging a pull request may close this issue.

None yet
3 participants
@VitalyFedyunin @f2013519 @gchanan
You can’t perform that action at this time.