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.

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

Colab compatibility improvements #87

Merged
merged 3 commits into from
Jun 19, 2020
Merged

Colab compatibility improvements #87

merged 3 commits into from
Jun 19, 2020

Conversation

AlphaGit
Copy link
Contributor

@AlphaGit AlphaGit commented Jun 7, 2020

Hi! I just added these fixes that improve on the Colab compatibility experience.

  • Ensure Google Colab GPU Compatibility
    • Added parameters to compile for all available GPU models in Google Colaboratory.
    • Moved the instructions to a common file used by all setup files
    • Adapted Colab file, removing outdated error indicators
  • Suppressing compiler warnings
  • Updated documentation
    • Updated documentation inside and outside of Colab file.
    • Removed references to any shared Colab file, to avoid outdated versions out there. The one in the master branch of this repo should be the canonical one.
    • Added versioning to the file itself, so we can easily tell when someone is using an old version of the file.
    • Added markdown input for configuration

- Added parameters to compile for all available GPU models in Google Colaboratory.
- Moved the instructions to a common file used by all setup files
- Adapted Colab file, removing outdated error indicators
- Updated documentation inside and outside of Colab file.
- Removed references to any shared Colab file, to avoid outdated versions out there. The one in the `master` branch of this repo should be the canonical one.
- Added versioning to the file itself, so we can easily tell when someone is using an old version of the file.
- Added markdown input for configuration
@baowenbo baowenbo merged commit 0e38076 into baowenbo:master Jun 19, 2020
@tianchengdw
Copy link

tianchengdw commented Sep 4, 2020

Hi, thank you for your commits, but I meet an error(error in correlation_forward_cuda_kernel: no kernel image is available for execution on the device) when I use Tesla V100 with compute capability 7.0.
I change the arch gen-code in the compiler_args.py. However, it seems that it doesn't work.
Thanks in advance!

@AlphaGit
Copy link
Contributor Author

AlphaGit commented Sep 4, 2020

Hi there! It seems like you're on the right path. Did you try adding both the arch and code to 70 compatibility?

Also, I know that this error can also occur on cases where the CUDA drivers are not quite supported -- I'm not entirely sure how to verify that's the case. Around the internet people just keep trying different versions of pytorch, cuda and nvidia drivers until they find a combination that works. What we did is fixed those in Colab to a known working version.

@tianchengdw
Copy link

Hi,thank you for your reply!

Yes,I use both the arch and code. First, I fork this repository from your Github. Then I add the
'-gencode', 'arch=compute_70,code=[sm_70,compute_70]' in the compiler_args.py. At last, I modify the git source in the colab file from my github. But it seems that it is not a useful way.

@emilwallner
Copy link

I tried it with a V100 on my Colab Pro account, and I had the same error. When I used a non-Colab Pro account, I got a T4 GPU and it worked as expected.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

4 participants