Skip to content
This repository was archived by the owner on Mar 28, 2023. It is now read-only.

[SYCL] Add a test for group sorting algorithm #438

Merged
merged 6 commits into from
Sep 20, 2021

Conversation

andreyfe1
Copy link

@andreyfe1 andreyfe1 commented Sep 2, 2021

Implementation of sorting algorithm is here: intel/llvm#4439

Cases that were added:
Data types:

  • uint32_t
  • double
  • sycl::half
  • uint16_t
  • size_t
  • int8_t
  • int32_t
  • float
  • custom type
    Functors:
  • less
  • greater
  • custom
    Sorters:
  • default
  • custom
    Sorting function:
  • joint
  • over_group
    Interfaces
  • sorter
  • compare
  • w/o compare or sorter
    Data size:
  • 1 group
  • multiple groups
    Generic group:
  • work-group

Cases that were not added:

  • sub-group

Signed-off-by: Fedorov, Andrey [email protected]

Signed-off-by: Fedorov, Andrey <[email protected]>
Signed-off-by: Fedorov, Andrey <[email protected]>
@andreyfe1
Copy link
Author

Inviting @romanovvlad for review. Feel free to invite more people if needed

Signed-off-by: Fedorov, Andrey <[email protected]>
Signed-off-by: Fedorov, Andrey <[email protected]>
@andreyfe1
Copy link
Author

Hi, all. Feel free to approve and merge if you don't have any more comments to the patch

@romanovvlad
Copy link

Hi, all. Feel free to approve and merge if you don't have any more comments to the patch

Currently the test fails in pre-commit - github UI doesn't allow to merge the patch. The implementation of the API this test validates should be merged first.

Signed-off-by: Fedorov, Andrey <[email protected]>
romanovvlad pushed a commit to intel/llvm that referenced this pull request Sep 15, 2021
…4439)

This PR introduce the implementation for the Group Sort extension.

The PR includes:

    Feature macro
    joint_sort and sort_over_group functions
    default_sorter
    group_with_scratchpad

Algorithms are quite general. It should work with custom data types, custom comparators, custom sorters.

The PR doesn't include:

    radix_sorter
    optimized specialization for arithmetic types

Tests are here: intel/llvm-test-suite#438

Signed-off-by: Fedorov, Andrey [email protected]
@andreyfe1
Copy link
Author

@romanovvlad, implementation was merged to the intel/llvm repo. CI is passed except Windows testing. Seems, failures are not related to the changes in this patch. What are the next step for this PR? Waiting until fix for Windows is not merged separately?

@andreyfe1
Copy link
Author

@Pennycook, do you have any other comments/questions related to the PR?

@romanovvlad romanovvlad merged commit a8567cf into intel:intel Sep 20, 2021
aelovikov-intel pushed a commit to aelovikov-intel/llvm that referenced this pull request Mar 27, 2023
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.

4 participants