KEMBAR78
[MPS] Fix index_add for complex + int64 by malfet · Pull Request #160926 · pytorch/pytorch · GitHub
Skip to content

[MPS] Fix index_add for complex + int64 #160926

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
wants to merge 1 commit into from
Closed

[MPS] Fix index_add for complex + int64 #160926

wants to merge 1 commit into from

Conversation

@malfet
Copy link
Contributor

@malfet malfet commented Aug 19, 2025 •
edited
Loading

Stack from ghstack (oldest at bottom):

By re-using deterministic algorithm from

pytorch/aten/src/ATen/native/cuda/Indexing.cu

Lines 1106 to 1113 in bbc7c03

if (globalContext().deterministicAlgorithms()){
torch::List<std::optional<Tensor>> indices;
indices.reserve(dim + 1);
for (const auto i: c10::irange(dim)) {
indices.emplace_back();
}
indices.emplace_back(index.to(at::kLong));
result.index_put_(indices, source * alpha, true);

Fixes #160845

@malfet malfet requested a review from kulinseth as a code owner August 19, 2025 00:19
@pytorch-bot
Copy link

pytorch-bot bot commented Aug 19, 2025 •
edited
Loading

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/160926

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

⏳ No Failures, 1 Pending

As of commit 87a151a with merge base 114813c (image):
💚 Looks good so far! There are no failures yet. 💚

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

malfet added a commit that referenced this pull request Aug 19, 2025
@malfet
[MPS] Fix index_add for complex + int64
a453f6a 
By re-using deterministic algorithm from
https://github.com/pytorch/pytorch/blob/bbc7c03e936ab0fce69dfda1fdf4798523a2fbf8/aten/src/ATen/native/cuda/Indexing.cu#L1107-L1111

ghstack-source-id: a13719f
Pull Request resolved: #160926
@pytorch-bot pytorch-bot bot added ciflow/mps Run MPS tests (subset of trunk) release notes: mps Release notes category labels Aug 19, 2025
@malfet malfet added topic: bug fixes topic category ciflow/trunk Trigger trunk jobs on your pull request labels Aug 19, 2025
@malfet
Copy link
Contributor Author

malfet commented Aug 19, 2025

@pytorchbot merge -f "Everything is green"

@pytorchmergebot
Copy link
Collaborator

Merge started

Your change will be merged immediately since you used the force (-f) flag, bypassing any CI checks (ETA: 1-5 minutes). Please use -f as last resort and instead consider -i/--ignore-current to continue the merge ignoring current failures. This will allow currently pending tests to finish and report signal before the merge.

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

@malfet malfet mentioned this pull request Aug 19, 2025
Closed
pytorchmergebot pushed a commit that referenced this pull request Aug 21, 2025
@malfet @pytorchmergebot
[BE] Move indexing tests to test_indexing (#160994)
3e3e834 
Which enables them on MPS device
- xfail all `test_index_reduce` on MPS, as op is not implemented
- xfail all `test_index_copy` on MPS due to the silent correctness problems, see #160993
- Fixed hard crash in `index_fill` and replaced `skipIfMPS` with `expectedFailueMPS`
- Created issue for the lack of deterministic algorithms for MPS backend
Pull Request resolved: #160994
Approved by: https://github.com/manuelcandales
ghstack dependencies: #160850, #160889, #160926
markc-614 pushed a commit to markc-614/pytorch that referenced this pull request Sep 17, 2025
@malfet @markc-614
[MPS] Fix index_add for complex + int64 (pytorch#160926)
66ef08c 
By re-using deterministic algorithm from
https://github.com/pytorch/pytorch/blob/bbc7c03e936ab0fce69dfda1fdf4798523a2fbf8/aten/src/ATen/native/cuda/Indexing.cu#L1106-L1113

Fixes pytorch#160845
Pull Request resolved: pytorch#160926
Approved by: https://github.com/manuelcandales
ghstack dependencies: pytorch#160850, pytorch#160889
markc-614 pushed a commit to markc-614/pytorch that referenced this pull request Sep 17, 2025
@malfet @markc-614
[BE] Move indexing tests to test_indexing (pytorch#160994)
9db3e87 
Which enables them on MPS device
- xfail all `test_index_reduce` on MPS, as op is not implemented
- xfail all `test_index_copy` on MPS due to the silent correctness problems, see pytorch#160993
- Fixed hard crash in `index_fill` and replaced `skipIfMPS` with `expectedFailueMPS`
- Created issue for the lack of deterministic algorithms for MPS backend
Pull Request resolved: pytorch#160994
Approved by: https://github.com/manuelcandales
ghstack dependencies: pytorch#160850, pytorch#160889, pytorch#160926
@github-actions github-actions bot deleted the gh/malfet/494/head branch September 19, 2025 02:10
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@manuelcandales manuelcandales manuelcandales approved these changes

@kulinseth kulinseth Awaiting requested review from kulinseth

@Skylion007 Skylion007 Awaiting requested review from Skylion007

@dcci dcci Awaiting requested review from dcci

Assignees

No one assigned

Labels

ciflow/mps Run MPS tests (subset of trunk) ciflow/trunk Trigger trunk jobs on your pull request Merged release notes: mps Release notes category topic: bug fixes topic category

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

3 participants

@malfet @pytorchmergebot @manuelcandales