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

Add gpu benchmark #330

Merged
merged 23 commits into from
Dec 13, 2023
Merged

Add gpu benchmark #330

merged 23 commits into from
Dec 13, 2023

Conversation

samkellerhals
Copy link
Contributor

@samkellerhals samkellerhals commented Dec 7, 2023 •
edited
Loading

Description

This PR updates the base image to support CUDA 11 for GPU backend testing.

Highlights:

  • Updated Base Image: Now supports CUDA 11, enabling GPU backend tests.
  • Benchmark Testing: Integrated into the PR CI pipeline, planning to shift to daily CI runs in the future.

Other Changes:

  • Moved CI runner to Daint.
  • Updated flake8.
  • Benchmark iterations reduced to 1.

@samkellerhals
Copy link
Contributor Author

cscs-ci run

@samkellerhals
Copy link
Contributor Author

cscs-ci run

1 similar comment
@samkellerhals
Copy link
Contributor Author

cscs-ci run

@samkellerhals
Copy link
Contributor Author

cscs-ci run

2 similar comments
@samkellerhals
Copy link
Contributor Author

cscs-ci run

@samkellerhals
Copy link
Contributor Author

cscs-ci run

@samkellerhals
Copy link
Contributor Author

launch jenkins spack

@samkellerhals
Copy link
Contributor Author

cscs-ci run

@samkellerhals
Copy link
Contributor Author

launch jenkins spack

@edopao
Copy link
Collaborator

edopao commented Dec 12, 2023

I agree with @DropD's comments, I have nothing to add on the code change.

It is not related about this PR, but I am concerned about the number figures. The test job on GPU node takes 51 minutes, and the benchmark results show execution time for each stencil in the order of 10 seconds:

test_TestMoSolveNonhydroStencil15_benchmark[backend=gtfn_gpu-grid=simple_grid] 10.4457 (1.09) 10.4457 (1.09) 10.4457 (1.09)

It is not much better on CPU, although a bit faster.

Thanks to this PR we will be able to track the variations (hopefully for the better) on the different kinds of node.

samkellerhals and others added 4 commits December 12, 2023 14:20
@samkellerhals @DropD
Update model/common/src/icon4py/model/common/test_utils/pytest_config.py
243dca8 
Co-authored-by: Rico Haeuselmann <r.haeuselmann@gmx.ch>
@samkellerhals @DropD
Update model/common/src/icon4py/model/common/test_utils/pytest_config.py
c9f1c5e 
Co-authored-by: Rico Haeuselmann <r.haeuselmann@gmx.ch>
@samkellerhals @DropD
Update model/common/src/icon4py/model/common/test_utils/pytest_config.py
c93f828 
Co-authored-by: Rico Haeuselmann <r.haeuselmann@gmx.ch>
@samkellerhals @DropD
Update model/common/src/icon4py/model/common/test_utils/pytest_config.py
cdc9fea 
Co-authored-by: Rico Haeuselmann <r.haeuselmann@gmx.ch>
@samkellerhals
Copy link
Contributor Author

cscs-ci run

@samkellerhals
Copy link
Contributor Author

cscs-ci run

@github-actions GitHub Actions
Copy link

Mandatory Tests

Please make sure you run these tests via comment before you merge!

  • cscs-ci run
  • launch jenkins spack

Optional Tests

In case your change might affect downstream icon-exclaim, please consider running

  • launch jenkins icon

For more detailed information please look at CI in the EXCLAIM universe.

@samkellerhals
Copy link
Contributor Author

cscs-ci run

1 similar comment
@samkellerhals
Copy link
Contributor Author

cscs-ci run

@samkellerhals
Copy link
Contributor Author

launch jenkins spack

@samkellerhals samkellerhals merged commit 46f6a5f into main Dec 13, 2023
5 checks passed
@samkellerhals samkellerhals deleted the add-gpu-benchmark branch December 13, 2023 09:05
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@DropD DropD DropD approved these changes

@edopao edopao Awaiting requested review from edopao

Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
3 participants
@samkellerhals @edopao @DropD