Skip to content

Implement cusparse Descriptor class and clean up cusparse code#37389

Closed
xwang233 wants to merge 9 commits intopytorch:masterfrom
xwang233:cusparse-descriptor
Closed

Implement cusparse Descriptor class and clean up cusparse code#37389
xwang233 wants to merge 9 commits intopytorch:masterfrom
xwang233:cusparse-descriptor

Conversation

@xwang233
Copy link
Copy Markdown
Collaborator

@xwang233 xwang233 commented Apr 28, 2020

Add cusparse Descriptor class. Add cusparse Generic API wrapper. Clean up current cuda sparse code.

@dr-ci
Copy link
Copy Markdown

dr-ci Bot commented Apr 28, 2020

💊 Build failures summary and remediations

As of commit bebc0dc (more details on the Dr. CI page):


  • 1/1 failures possibly* introduced in this PR
    • 1/1 non-CircleCI failure(s)

ci.pytorch.org: 1 failed


This comment was automatically generated by Dr. CI (expand for details).Follow this link to opt-out of these comments for your Pull Requests.

Please report bugs/suggestions on the GitHub issue tracker.

See how this bot performed.

This comment has been revised 14 times.

@xwang233 xwang233 changed the title [WIP] Implement Descriptor class to simplify cusparse code [WIP] Implement cusparse Descriptor class and clean up cusparse code Apr 28, 2020
@xwang233 xwang233 changed the title [WIP] Implement cusparse Descriptor class and clean up cusparse code Implement cusparse Descriptor class and clean up cusparse code Apr 28, 2020
@xwang233 xwang233 requested review from ezyang and kurtamohler April 28, 2020 19:04
@xwang233
Copy link
Copy Markdown
Collaborator Author

cc @csarofeen @ptrblck

#pragma once

// guard the whole file
#if !defined(_MSC_VER) && defined(__CUDACC__) && CUSPARSE_VERSION >= 10301 // CUDA release >= 10.2 and not windows
Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

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

Shouldn't this be CUSPARSE_VERSION >= 10200 for v10.2?

Copy link
Copy Markdown
Collaborator Author

Choose a reason for hiding this comment

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

Hmm, not sure why, but I checked my header file and it's cusparse_version 10301 for my cuda 10.2.89. Also, there was a confirmed mismatch for cusparseSpMM on cuda 10.1

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

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

Oh you're right. I have the same version of cuda and I see 10301 too.

Copy link
Copy Markdown
Collaborator

@kurtamohler kurtamohler left a comment

Choose a reason for hiding this comment

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

Nice work! BTW, I don't think the pr/py3.6-clang7-rocmdeb-ubuntu16.04 CI failure is caused by your changes, because I've seen the same thing in a different PR.

Copy link
Copy Markdown
Contributor

@facebook-github-bot facebook-github-bot left a comment

Choose a reason for hiding this comment

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

@ezyang is landing this pull request. If you are a Facebook employee, you can view this diff on Phabricator.

@facebook-github-bot
Copy link
Copy Markdown
Contributor

@ezyang merged this pull request in bbf29a5.

facebook-github-bot pushed a commit that referenced this pull request Apr 30, 2020
…de (#37533)

Summary:
Reland of #37389. Fix for the cuda 10.1 CI failure.
Pull Request resolved: #37533

Differential Revision: D21326386

Pulled By: ezyang

fbshipit-source-id: f34875865c8bad76163995c18d88b0e76656bb22
laurentdupin pushed a commit to laurentdupin/pytorch that referenced this pull request Apr 24, 2026
…ch#37389)

Summary:
Add cusparse Descriptor class. Add cusparse Generic API wrapper. Clean up current cuda sparse code.
Pull Request resolved: pytorch#37389

Differential Revision: D21302691

Pulled By: ezyang

fbshipit-source-id: ecbb4063466c616eebfe681f1622724692be505c
laurentdupin pushed a commit to laurentdupin/pytorch that referenced this pull request Apr 24, 2026
…de (pytorch#37533)

Summary:
Reland of pytorch#37389. Fix for the cuda 10.1 CI failure.
Pull Request resolved: pytorch#37533

Differential Revision: D21326386

Pulled By: ezyang

fbshipit-source-id: f34875865c8bad76163995c18d88b0e76656bb22
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

6 participants