KEMBAR78
[CUDA] fix nansum in non-JIT build by pzzp · Pull Request #158633 · pytorch/pytorch · GitHub
Skip to content

[CUDA] fix nansum in non-JIT build #158633

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

[CUDA] fix nansum in non-JIT build #158633

wants to merge 1 commit into from

Conversation

@pzzp
Copy link
Contributor

@pzzp pzzp commented Jul 18, 2025

This change fix crash of

import torch
a = torch.tensor([[1, 2]], dtype=torch.complex32).to('cuda')
b = torch.nansum(a, dim=0)
print(b)

@pzzp
[CUDA] fix nansum in non-JIT build
7eb797b 
This change fix crash of
```
import torch
a = torch.tensor([[1, 2]], dtype=torch.complex32).to('cuda')
b = torch.nansum(a, dim=0)
print(b)
```
@pzzp pzzp requested review from eqy and syed-ahmed as code owners July 18, 2025 09:57
@pytorch-bot
Copy link

pytorch-bot bot commented Jul 18, 2025 •
edited
Loading

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/158633

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

✅ You can merge normally! (1 Unrelated Failure)

As of commit 7eb797b with merge base 32aade9 (image):

UNSTABLE - The following job is marked as unstable, possibly due to flakiness on trunk:

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

@linux-foundation-easycla
Copy link

linux-foundation-easycla bot commented Jul 18, 2025 •
edited
Loading

CLA Signed


The committers listed above are authorized under a signed CLA.

@pytorch-bot pytorch-bot bot added the release notes: cuda release notes category label Jul 18, 2025
@pzzp pzzp mentioned this pull request Jul 18, 2025
Closed
@malfet
Copy link
Contributor

malfet commented Jul 18, 2025

@pzzp can you sign the CLA please?

@pzzp
Copy link
Contributor Author

pzzp commented Jul 18, 2025

@malfet ✅️

@thenumberouscode
Copy link
Contributor

Hi, I’m just curious why changing acc_t to scalar_t fixes this bug.

@pzzp
Copy link
Contributor Author

pzzp commented Jul 21, 2025

@thenumberouscode
The correct computation process is: input scalar_t (complex half), reduce using the acc_t (complex float) type, and output out_scalar_t (complex half). This bug occurred because the output_scalar_t type was incorrect, leading to miscalculated output size and resulting in unaligned memory access.

image

@pzzp
Copy link
Contributor Author

pzzp commented Jul 27, 2025

@ngimel hi, how can I merge this change?

@ngimel
Copy link
Collaborator

ngimel commented Jul 28, 2025

@pytorchbot merge

@pytorch-bot pytorch-bot bot added the ciflow/trunk Trigger trunk jobs on your pull request label Jul 28, 2025
@pytorchmergebot
Copy link
Collaborator

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

yangw-dev pushed a commit that referenced this pull request Aug 1, 2025
@pzzp @yangw-dev
[CUDA] fix nansum in non-JIT build (#158633)
a6de1f7 
This change fix crash of
```
import torch
a = torch.tensor([[1, 2]], dtype=torch.complex32).to('cuda')
b = torch.nansum(a, dim=0)
print(b)
```

Pull Request resolved: #158633
Approved by: https://github.com/ngimel
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@ngimel ngimel ngimel approved these changes

@eqy eqy Awaiting requested review from eqy eqy is a code owner

@syed-ahmed syed-ahmed Awaiting requested review from syed-ahmed syed-ahmed is a code owner

Assignees

No one assigned

Labels

ciflow/trunk Trigger trunk jobs on your pull request Merged open source release notes: cuda release notes category

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

6 participants

@pzzp @malfet @thenumberouscode @ngimel @pytorchmergebot @pytorchbot