KEMBAR78
gen ExclusivelyOwned in structured kernels by swolchok · Pull Request #59827 · pytorch/pytorch · GitHub
Skip to content

Conversation

@swolchok
Copy link
Contributor

@swolchok swolchok commented Jun 10, 2021

@facebook-github-bot
Copy link
Contributor

facebook-github-bot commented Jun 10, 2021

💊 CI failures summary and remediations

As of commit eeb7ed5 (more details on the Dr. CI page and at hud.pytorch.org/pr/59827):


  • 1/2 failures possibly* introduced in this PR
    • 1/1 non-scanned failure(s)
  • 1/2 broken upstream at merge base a88e959 from Jun 09 until Jun 11

🚧 1 fixed upstream failure:

These were probably caused by upstream breakages that were already fixed.

Please rebase on the viable/strict branch (expand for instructions)

If your commit is older than viable/strict, run these commands:

git fetch https://github.com/pytorch/pytorch viable/strict
git rebase FETCH_HEAD

Preview docs built from this PR

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 to the (internal) Dr. CI Users group.

Click here to manually regenerate this comment.

swolchok added a commit that referenced this pull request Jun 10, 2021
Differential Revision: [D28965922](https://our.internmc.facebook.com/intern/diff/D28965922/)

ghstack-source-id: 131122290
Pull Request resolved: #59827
@swolchok swolchok requested review from bhosmer and ezyang June 15, 2021 17:11
@ezyang
Copy link
Contributor

ezyang commented Jun 15, 2021

Could you post the before and after generated code?

@ezyang
Copy link
Contributor

ezyang commented Jun 15, 2021

It doesn't feel like, as is, there is any benefit to ExclusivelyOwned here, yet. Is that true?

@ezyang ezyang closed this Jun 15, 2021
@ezyang ezyang reopened this Jun 15, 2021
@swolchok
Copy link
Contributor Author

It doesn't feel like, as is, there is any benefit to ExclusivelyOwned here, yet. Is that true?

It should save us a refcount decrement for the functional operators.

@swolchok
Copy link
Contributor Author

It should save us a refcount decrement for the functional operators.

Correction, it should save us a ~Tensor() call, which would have to do a load and branch to check if the TensorImpl was UndefinedTensorImpl. It would always ffind that to be the case and thus wouldn't do a refcount decrement anyway, though. I put pastes with the before/after RegisterCPU.cpp on Phabricator.

swolchok added a commit that referenced this pull request Jun 18, 2021
Pull Request resolved: #59827


ghstack-source-id: 131793841

Differential Revision: [D28965922](https://our.internmc.facebook.com/intern/diff/D28965922/)
@facebook-github-bot
Copy link
Contributor

This pull request has been merged in a5c5b56.

@facebook-github-bot facebook-github-bot deleted the gh/swolchok/255/head branch July 13, 2021 14:17
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.

4 participants