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This repository has been archived by the owner on Nov 17, 2023.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

[Operator] contrib.index_array op improvements #16639

Open
wants to merge 10 commits into
base: master
Choose a base branch
from

Conversation

nickguletskii
Copy link
Contributor

Description

Adds large tensor support to contrib.index_array, changes the implementation to use cudaMemcpyAsync (#16583), adds a parameter called target_axis that makes it easier to use this operator together with operators which expect the index dimension to be the first (such as gather_nd), adds a definition for FIgnoreInputs, and adds a dtype parameter.

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

  • Large tensor support, with nightly test.
  • Use cudaMemcpyAsync instead of cudaMemcpy.
  • Add a target_axis parameter that changes the axis for the indices.
  • Add FIgnoreInputs.
  • Add a dtype parameter.

Comments

@nickguletskii
Copy link
Contributor Author

The CI failures are due to the flaky np.einsum test (#16616).

Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

1 participant