Skip to content

[Test][Easy] Use float16 dtype in test_sort_large #159939

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
Closed
Aidyn-A wants to merge 1 commit into from
Closed

[Test][Easy] Use float16 dtype in test_sort_large #159939

Aidyn-A wants to merge 1 commit into from

Conversation

@Aidyn-A
Copy link
Collaborator

@Aidyn-A Aidyn-A commented Aug 6, 2025 •
edited by pytorch-bot bot
Loading

The test fails with:

RuntimeError: var_mean only support floating point and complex dtypes

cc @ptrblck @msaroufim @eqy @jerryzh168

@Aidyn-A Aidyn-A requested a review from eqy August 6, 2025 06:25
@Aidyn-A Aidyn-A self-assigned this Aug 6, 2025
@Aidyn-A Aidyn-A added module: cuda Related to torch.cuda, and CUDA support in general topic: not user facing topic category labels Aug 6, 2025
@pytorch-bot
Copy link

pytorch-bot bot commented Aug 6, 2025 •
edited
Loading

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/159939

Note: Links to docs will display an error until the docs builds have been completed.

✅ No Failures

As of commit 712d30f with merge base a53d14d (image):
💚 Looks good so far! There are no failures yet. 💚

This comment was automatically generated by Dr. CI and updates every 15 minutes.

eqy
eqy approved these changes Aug 6, 2025
View reviewed changes
Copy link
Collaborator

@eqy eqy left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

approving assuming that this never ran before and therefore this is not a regression

@Aidyn-A
Copy link
Collaborator Author

Aidyn-A commented Aug 6, 2025

approving assuming that this never ran before and therefore this is not a regression

Indeed, it never ran before. The only machine which has that much memory is GB300.

@Aidyn-A
Copy link
Collaborator Author

Aidyn-A commented Aug 7, 2025

Hmm, those are strange failures

@pytorchbot rebase

@pytorchmergebot
Copy link
Collaborator

pytorchmergebot commented Aug 7, 2025

@pytorchbot started a rebase job onto refs/remotes/origin/viable/strict. Check the current status here

@pytorchmergebot
Copy link
Collaborator

pytorchmergebot commented Aug 7, 2025

Successfully rebased test_sort_large_float16 onto refs/remotes/origin/viable/strict, please pull locally before adding more changes (for example, via git checkout test_sort_large_float16 && git pull --rebase)

@pytorchmergebot pytorchmergebot force-pushed the test_sort_large_float16 branch from a321cf0 to 712d30f Compare August 7, 2025 17:41
@Aidyn-A Aidyn-A added the ciflow/trunk Trigger trunk jobs on your pull request label Aug 7, 2025
@Aidyn-A
Copy link
Collaborator Author

Aidyn-A commented Aug 8, 2025

@pytorchbot merge

@pytorchmergebot
Copy link
Collaborator

pytorchmergebot commented Aug 8, 2025

Merge started

Your change will be merged once all checks pass (ETA 0-4 Hours).

Learn more about merging in the wiki.

Questions? Feedback? Please reach out to the PyTorch DevX Team

Advanced Debugging
Check the merge workflow status
here

@jithunnair-amd jithunnair-amd mentioned this pull request Aug 8, 2025
Closed
@jithunnair-amd
Copy link
Collaborator

jithunnair-amd commented Aug 8, 2025

approving assuming that this never ran before and therefore this is not a regression

Indeed, it never ran before. The only machine which has that much memory is GB300.

@Aidyn-A Did this updated test run in any of the CI jobs? It failed in ROCm CI (because the MI325 has >200GB memory) with the error Cannot sort dimension of length 8192 (link) (error coming from here), but I can't tell if it passed for CUDA at all.

@Aidyn-A
Copy link
Collaborator Author

Aidyn-A commented Aug 8, 2025

approving assuming that this never ran before and therefore this is not a regression

Indeed, it never ran before. The only machine which has that much memory is GB300.

@Aidyn-A Did this updated test run in any of the CI jobs? It failed in ROCm CI (because the MI325 has >200GB memory) with the error Cannot sort dimension of length 8192 (link) (error coming from here), but I can't tell if it passed for CUDA at all.

Yes, it is passing on GB300. Was it passing prior my changes? I believe it should have failed, because var_mean does not support integer dtypes.

hinriksnaer pushed a commit to hinriksnaer/pytorch that referenced this pull request Aug 8, 2025
@Aidyn-A @hinriksnaer
[Test][Easy] Use float16 dtype in test_sort_large (pytorch#159939)
016351c 
The test fails with:
>RuntimeError: var_mean only support floating point and complex dtypes

Pull Request resolved: pytorch#159939
Approved by: https://github.com/eqy
BLOrange-AMD pushed a commit to ROCm/pytorch that referenced this pull request Aug 19, 2025
@Aidyn-A @BLOrange-AMD
[Test][Easy] Use float16 dtype in test_sort_large (pytorch#159939)
9fddb06 
The test fails with:
>RuntimeError: var_mean only support floating point and complex dtypes

Pull Request resolved: pytorch#159939
Approved by: https://github.com/eqy
@dnikolaev-amd dnikolaev-amd mentioned this pull request Aug 21, 2025
Merged
jithunnair-amd pushed a commit to ROCm/pytorch that referenced this pull request Aug 25, 2025
@dnikolaev-amd
[rocm7.1_internal_testing] fix large tensor sort on ROCm (#2543)
1455054 
Currently std::min -> ::min did not work as expected on ROCm when input
values >= 2147483648

Replace std::min to ternary statement
Also std::min can be replaced by explicit typing std::min<int64_t>

fixes on ROCm:

test_sort_and_select.py::TestSortAndSelectCUDA::test_sort_large_cuda_float16
error:
RuntimeError: Cannot sort dimension of length 8192

Combines upstream PRs:
- pytorch#161054 to fix std::min on ROCm
- pytorch#155546 fix python test
- pytorch#159939 change test dtype from
int8 to float16

Fixes: SWDEV-526432
dhonnappa-amd pushed a commit to ROCm/pytorch that referenced this pull request Aug 25, 2025
@dnikolaev-amd
[rocm7.1_internal_testing] fix large tensor sort on ROCm (#2543)
50ddaf7 
Currently std::min -> ::min did not work as expected on ROCm when input
values >= 2147483648

Replace std::min to ternary statement
Also std::min can be replaced by explicit typing std::min<int64_t>

fixes on ROCm:

test_sort_and_select.py::TestSortAndSelectCUDA::test_sort_large_cuda_float16
error:
RuntimeError: Cannot sort dimension of length 8192

Combines upstream PRs:
- pytorch#161054 to fix std::min on ROCm
- pytorch#155546 fix python test
- pytorch#159939 change test dtype from
int8 to float16

Fixes: SWDEV-526432
dhonnappa-amd pushed a commit to ROCm/pytorch that referenced this pull request Aug 25, 2025
@dnikolaev-amd
[rocm7.1_internal_testing] fix large tensor sort on ROCm (#2543)
d57ba11 
Currently std::min -> ::min did not work as expected on ROCm when input
values >= 2147483648

Replace std::min to ternary statement
Also std::min can be replaced by explicit typing std::min<int64_t>

fixes on ROCm:

test_sort_and_select.py::TestSortAndSelectCUDA::test_sort_large_cuda_float16
error:
RuntimeError: Cannot sort dimension of length 8192

Combines upstream PRs:
- pytorch#161054 to fix std::min on ROCm
- pytorch#155546 fix python test
- pytorch#159939 change test dtype from
int8 to float16

Fixes: SWDEV-526432
markc-614 pushed a commit to markc-614/pytorch that referenced this pull request Sep 17, 2025
@Aidyn-A @markc-614
[Test][Easy] Use float16 dtype in test_sort_large (pytorch#159939)
1b5d765 
The test fails with:
>RuntimeError: var_mean only support floating point and complex dtypes

Pull Request resolved: pytorch#159939
Approved by: https://github.com/eqy
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@eqy eqy eqy approved these changes

Assignees

@Aidyn-A Aidyn-A

Labels

ciflow/trunk Trigger trunk jobs on your pull request Merged module: cuda Related to torch.cuda, and CUDA support in general open source topic: not user facing topic category

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

5 participants

@Aidyn-A @pytorchmergebot @jithunnair-amd @eqy @pytorchbot