Skip to content

Enable simd and loop vectorizer with MSVC #20530

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 6 commits into from
Closed

Enable simd and loop vectorizer with MSVC #20530

wants to merge 6 commits into from

Conversation

peterjc123
Copy link
Collaborator

No description provided.

@pytorchbot pytorchbot added module: ci Related to continuous integration module: cpu CPU specific problem (e.g., perf, algorithm) module: operators labels May 15, 2019
@peterjc123 peterjc123 changed the title Fix simd and loop vectorizer with MSVC [WIP] Fix simd and loop vectorizer with MSVC May 15, 2019
@peterjc123
Copy link
Collaborator Author

From the log, it seems that it is a long way to go. https://ci.pytorch.org/jenkins/job/pytorch-builds/job/pytorch-win-ws2016-cuda10-cudnn7-py3-dev-build/3/consoleText

@peterjc123
Copy link
Collaborator Author

The reason codes for vectorization failures: https://docs.microsoft.com/en-us/cpp/error-messages/tool-errors/vectorizer-and-parallelizer-messages?view=vs-2019

@peterjc123 peterjc123 mentioned this pull request May 15, 2019
Closed
@peterjc123
Copy link
Collaborator Author

@pytorchbot rebase this please

@pytorchbot pytorchbot added caffe2 module: build Build system issues labels May 16, 2019
@peterjc123
Copy link
Collaborator Author

peterjc123 commented May 16, 2019 •
edited
Loading

The auto-vectorization report for VS 2017: https://gist.github.com/peterjc123/714c40634ceeb9c5a533e8d78b5f24c4.

The auto-vectorization report for VS 2019:
https://gist.github.com/peterjc123/019b084e7f15f2b372a662f6ce92f786.

@peterjc123 peterjc123 added module: openmp Related to OpenMP (omp) support in PyTorch module: vectorization Related to SIMD vectorization, e.g., Vec256 labels May 16, 2019
@li-roy li-roy added the triaged This issue has been looked at a team member, and triaged and prioritized into an appropriate module label May 16, 2019
@peterjc123
Copy link
Collaborator Author

Let's enable it first. Fixing the non-vectorized loops could be done later.

@peterjc123 peterjc123 requested a review from ezyang May 17, 2019 07:09
@peterjc123 peterjc123 changed the title [WIP] Fix simd and loop vectorizer with MSVC Enable simd and loop vectorizer with MSVC May 17, 2019
facebook-github-bot
facebook-github-bot reviewed May 17, 2019
View reviewed changes
Copy link
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.

@peterjc123 peterjc123 deleted the vec_fix branch May 17, 2019 16:01
@peterjc123
Copy link
Collaborator Author

@ezyang Actually I hope someone that knows/manages the vectorization of TH/THNN may look into these vectorization reports because I'm not sure which ones of those functions should be vectorized and which ones are not.

@facebook-github-bot
Copy link
Contributor

@ezyang merged this pull request in 1136ad5.

zdevito pushed a commit to zdevito/ATen that referenced this pull request May 17, 2019
@peterjc123 @facebook-github-bot
Enable simd and loop vectorizer with MSVC
f3ffb48 
Summary: Pull Request resolved: pytorch/pytorch#20530

Differential Revision: D15392676

Pulled By: ezyang

fbshipit-source-id: c8fda0c7835127f81adf55016223bb4dc14ff40a
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@facebook-github-bot facebook-github-bot facebook-github-bot left review comments

@ezyang ezyang ezyang approved these changes

Assignees

No one assigned

Labels

caffe2 Merged module: build Build system issues module: ci Related to continuous integration module: cpu CPU specific problem (e.g., perf, algorithm) module: openmp Related to OpenMP (omp) support in PyTorch module: vectorization Related to SIMD vectorization, e.g., Vec256 open source triaged This issue has been looked at a team member, and triaged and prioritized into an appropriate module

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

6 participants

@peterjc123 @facebook-github-bot @ezyang @li-roy @pytorchbot @mruberry