Skip to content
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

Support segmented reductions and null mask reductions #9621

Merged
merged 67 commits into from
Mar 10, 2022
Merged

Support segmented reductions and null mask reductions #9621

merged 67 commits into from
Mar 10, 2022

Conversation

isVoid
Copy link
Contributor

@isVoid isVoid commented Nov 5, 2021 •
edited
Loading

closes #9135
closes #9552

This PR adds support for numeric types to simple_op, sum, prod, min, max, any, all. Also, this PR adds segmented_null_mask_reduction to compute null mask reductions on segments.

@isVoid isVoid requested a review from a team as a code owner November 5, 2021 21:54
@isVoid isVoid requested review from trxcllnt and jrhemstad November 5, 2021 21:54
@github-actions github-actions bot added CMake CMake build issue libcudf Affects libcudf (C++/CUDA) code. labels Nov 5, 2021
@isVoid isVoid marked this pull request as draft November 5, 2021 21:54
jrhemstad
jrhemstad requested changes Nov 6, 2021
View reviewed changes
@jrhemstad jrhemstad changed the title [WIP] Support segment reduce [WIP] Support segmented reductions Nov 10, 2021
@ttnghia
Copy link
Contributor

ttnghia commented Nov 15, 2021

I didn't read the implementation yet. But I assume that this is extremely similar to what sort-based groupby aggregate is doing. If this new implementation is more efficient than the current groupby aggregate implementation, we also need to switch to use segmented reduction for groupby aggregate too. In order to support this idea, please add a benchmark for comparison between their performance.

@github-actions github-actions bot added conda Java Affects Java cuDF API. Python Affects Python cuDF API. labels Dec 14, 2021
@isVoid isVoid changed the base branch from branch-21.12 to branch-22.02 December 14, 2021 22:00
@bdice bdice mentioned this pull request Dec 14, 2021
Merged
@github-actions github-actions bot removed gpuCI Python Affects Python cuDF API. labels Dec 15, 2021
@isVoid
Copy link
Contributor Author

isVoid commented Feb 17, 2022

I updated the docstring for segmented_reduction, to describe the specification of offsets. Also added a note on the current up-cast of precision when reducing integral and floating point values, and remove mentioning of computation precision at output_dtype.

revans2
revans2 approved these changes Feb 17, 2022
View reviewed changes
Copy link
Contributor

@revans2 revans2 left a comment

Choose a reason for hiding this comment

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

This combined with #10302 looks to give me what I need

rapids-bot bot pushed a commit that referenced this pull request Feb 17, 2022
@bdice
Add conversions between column_view and device_span<T const>. (#10302)
d48dd6f 
This PR adds an implicit conversion operator from `column_view` to `device_span<T const>`. The immediate purpose of this PR is to make it possible to use the API `segmented_reduce(column_view data, device_span<size_type> offsets, ...)` in PR #9621.

This PR also resolves #9656 by adding a `column_view` constructor from `device_span<T const>`.

More broadly, this PR should make it easier to refactor instances where `column.data()` is used with counting iterators to build transform iterators, or other patterns that require a length (e.g. vector factories to copy to host).

Authors:
  - Bradley Dice (https://github.com/bdice)

Approvers:
  - Mark Harris (https://github.com/harrism)
  - Jake Hemstad (https://github.com/jrhemstad)
  - David Wendt (https://github.com/davidwendt)

URL: #10302
bdice
bdice approved these changes Feb 17, 2022
View reviewed changes
Copy link
Contributor

@bdice bdice left a comment

Choose a reason for hiding this comment

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

I have a few minor comments but I think this is sufficiently ready for merge. We have a few TODO items that we should probably track in an issue.

@revans2 revans2 mentioned this pull request Feb 18, 2022
Closed
@isVoid isVoid requested a review from jrhemstad February 25, 2022 18:54
@bdice
Copy link
Contributor

bdice commented Mar 10, 2022 •
edited
Loading

@gpucibot merge Oops. This is blocked until @jrhemstad approves / retracts request for changes.

@bdice
Copy link
Contributor

bdice commented Mar 10, 2022

@gpucibot merge

@rapids-bot rapids-bot bot merged commit 7ff1956 into rapidsai:branch-22.04 Mar 10, 2022
This was referenced Mar 11, 2022
Merged
Closed
rapids-bot bot pushed a commit that referenced this pull request Mar 14, 2022
@revans2
JNI support for segmented reduce (#10413)
a066e7f 
This adds in JNI support for #9621. It also adds in a helper API to allow us to do the processing on a list easily.

Authors:
  - Robert (Bobby) Evans (https://github.com/revans2)

Approvers:
  - Liangcai Li (https://github.com/firestarman)
  - Jason Lowe (https://github.com/jlowe)

URL: #10413
@bdice bdice mentioned this pull request Mar 14, 2022
Open
@GregoryKimball GregoryKimball mentioned this pull request Mar 25, 2022
Closed
@bdice bdice mentioned this pull request Jun 15, 2022
Closed
@GregoryKimball GregoryKimball mentioned this pull request Jun 24, 2022
Open
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@vyasr vyasr vyasr approved these changes

@bdice bdice bdice approved these changes

@nvdbaranec nvdbaranec nvdbaranec left review comments

@revans2 revans2 revans2 approved these changes

@jrhemstad jrhemstad jrhemstad approved these changes

@trxcllnt trxcllnt Awaiting requested review from trxcllnt trxcllnt is a code owner automatically assigned from rapidsai/cudf-cpp-codeowners

Assignees

@bdice bdice

@isVoid isVoid

Labels
CMake CMake build issue feature request New feature or request libcudf Affects libcudf (C++/CUDA) code. non-breaking Non-breaking change
Projects
None yet
Milestone
No milestone
7 participants
@isVoid @ttnghia @vyasr @revans2 @bdice @jrhemstad @nvdbaranec