Skip to content

Add torch.backends.cudnn.is_available #8703

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
Merged
goldsborough merged 1 commit into from
Jun 20, 2018
Merged

Add torch.backends.cudnn.is_available #8703

goldsborough merged 1 commit into from
Jun 20, 2018

Conversation

@goldsborough
Copy link
Contributor

@goldsborough goldsborough commented Jun 20, 2018

Some users are using torch._C.has_cudnn to determine if cudnn is available (e.g. https://github.com/pytorch/pytorch/pull/8313/files#diff-52a3dcb763bfba3076c1cc8aef9b1e72R17). There should probably be a friendlier way of doing this. This PR adds this function.

Please advise if torch._C.has_cudnn is ok, or better torch.backends.cudnn.is_acceptable(torch.tensor(1., device=CUDA_DEVICE)) which I've seen in other code.

@soumith @apaszke @colesbury

@goldsborough goldsborough mentioned this pull request Jun 20, 2018
Merged
soumith
soumith reviewed Jun 20, 2018
View reviewed changes
test/common_cuda.py Outdated

This comment was marked as off-topic.

Sign in to view

This comment was marked as off-topic.

Sign in to view

@ssnl
Copy link
Collaborator

ssnl commented Jun 20, 2018 •
edited
Loading

Why is this under torch.cuda instead of torch.backends.cudnn? What's the general guideline here?

I'm asking because I'm adding some cuFFT specific methods in #8344 and want to know where they should live.

@apaszke
Copy link
Contributor

apaszke commented Jun 20, 2018

Yeah, why are we duplicating this? We have torch.backends.cudnn.is_available()

@goldsborough
Copy link
Contributor Author

goldsborough commented Jun 20, 2018

@apaszke I don't think this function you mention exists?

I don't mind where it lives, just let me know what you would prefer

@ssnl
Copy link
Collaborator

ssnl commented Jun 20, 2018

@apaszke we only have torch.backends.mkl.is_available(). IMO, it would be better for this to be torch.backends.cudnn.is_available().

@goldsborough
Copy link
Contributor Author

goldsborough commented Jun 20, 2018

done

@goldsborough goldsborough changed the title Add torch.cuda.cudnn_is_available Add torch.backends.cudnn.is_available Jun 20, 2018
@apaszke
Copy link
Contributor

apaszke commented Jun 20, 2018

Hmm ok, I could swear it exists 😕 Adding it sounds good to me

@goldsborough goldsborough merged commit 0acddd6 into pytorch:master Jun 20, 2018
petrex pushed a commit to petrex/pytorch that referenced this pull request Jun 21, 2018
@goldsborough
Add torch.cuda.cudnn_is_available (pytorch#8703)
2a5223b
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@apaszke apaszke Awaiting requested review from apaszke

@colesbury colesbury Awaiting requested review from colesbury

@ezyang ezyang Awaiting requested review from ezyang

@gchanan gchanan Awaiting requested review from gchanan

@zdevito zdevito Awaiting requested review from zdevito

1 more reviewer

@soumith soumith soumith approved these changes

Reviewers whose approvals may not affect merge requirements

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

4 participants

@goldsborough @ssnl @apaszke @soumith