Skip to content
This repository has been archived by the owner on Nov 17, 2023. It is now read-only.

[MKLDNN] Use MKLDNNRun #16772

Merged
merged 8 commits into from
Nov 29, 2019
Merged

Conversation

ZhennanQin
Copy link
Contributor

Description

MKLDNNRun will handle ndarray view and enhance code quality.

@PatricZhao @TaoLv

Checklist

Essentials

Please feel free to remove inapplicable items for your PR.

  • The PR title starts with [MXNET-$JIRA_ID], where $JIRA_ID refers to the relevant JIRA issue created (except PRs with tiny changes)
  • Changes are complete (i.e. I finished coding on this PR)
  • All changes have test coverage:
  • Unit tests are added for small changes to verify correctness (e.g. adding a new operator)
  • Nightly tests are added for complicated/long-running ones (e.g. changing distributed kvstore)
  • Build tests will be added for build configuration changes (e.g. adding a new build option with NCCL)
  • Code is well-documented:
  • For user-facing API changes, API doc string has been updated.
  • For new C++ functions in header files, their functionalities and arguments are documented.
  • For new examples, README.md is added to explain the what the example does, the source of the dataset, expected performance on test set and reference to the original paper if applicable
  • Check the API doc at https://mxnet-ci-doc.s3-accelerate.dualstack.amazonaws.com/PR-$PR_ID/$BUILD_ID/index.html
  • To the my best knowledge, examples are either not affected by this change, or have been fixed to be compatible with this change

Changes

  • Feature1, tests, (and when applicable, API doc)
  • Feature2, tests, (and when applicable, API doc)

Comments

  • If this change is a backward incompatible change, why must this change be made.
  • Interesting edge cases to note here

@pengzhao-intel
Copy link
Contributor

LGTM but let's wait for the performance testing reports.

@pengzhao-intel
Copy link
Contributor

@TaoLv @ciyongch any comments?

@pengzhao-intel pengzhao-intel added this to In progress in CPU Performance and Quantization via automation Nov 14, 2019
@TaoLv
Copy link
Member

TaoLv commented Nov 14, 2019

Do we fix any specific issue with these changes or they're just for code refactoring?

@ZhennanQin
Copy link
Contributor Author

Do we fix any specific issue with these changes or they're just for code refactoring?

Not for any particular issue, but a code refactoring to improve code quality. Currently, ndarray view check doesn't cover all mkldnn operators, which may cause trouble in rare case. This is to prevent that happen.

@pengzhao-intel
Copy link
Contributor

@TaoLv is this refactor OK?

Copy link
Contributor

@ciyongch ciyongch left a comment

Choose a reason for hiding this comment

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

Looks good :)
One question: MKLDNNRun is checking and doing reorder for an input, how about multi-inputs Ops like elemwise_add (MKLDNNSumForward), better to check all the inputs, right?

@ZhennanQin
Copy link
Contributor Author

@ciyongch Yes. MKLDNNRun has 2 versions. One is for unary ops, and other one is for ops with multiple inputs & outputs. elemwise_add will use the second version, and all inputs are well handled.

Copy link
Contributor

@ciyongch ciyongch left a comment

Choose a reason for hiding this comment

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

@ZhennanQin I see, I didn't realize that version was already in the master branch.

CPU Performance and Quantization automation moved this from In progress to Reviewer approved Nov 19, 2019
@pengzhao-intel
Copy link
Contributor

@ZhennanQin please rebase the code

@pengzhao-intel pengzhao-intel merged commit 5fb2916 into apache:master Nov 29, 2019
CPU Performance and Quantization automation moved this from Reviewer approved to Done Nov 29, 2019
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
Projects
No open projects
Development

Successfully merging this pull request may close these issues.

None yet

4 participants