Make device_buffer streams explicit and enforce move construction #8280
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
… of device_buffer (move instead), etc.
harrism
added
improvement
Codecov Report
@@ Coverage Diff @@
## branch-21.06 #8280 +/- ##
===============================================
Coverage ? 82.89%
===============================================
Files ? 105
Lines ? 17875
Branches ? 0
===============================================
Hits ? 14817
Misses ? 3058
Partials ? 0 Continue to review full report at Codecov.
|
jrhemstad
reviewed
May 19, 2021
jrhemstad
reviewed
May 19, 2021
jrhemstad
reviewed
May 19, 2021
rgsl888prabhu
approved these changes
May 19, 2021
devavret
approved these changes
May 20, 2021
ttnghia
approved these changes
May 20, 2021
Co-authored-by: Nghia Truong <ttnghia@users.noreply.github.com>
Co-authored-by: Nghia Truong <ttnghia@users.noreply.github.com>
4 tasks
harrism
added
5 - Ready to Merge
|
@gpucibot merge |
|
Rerun tests. |
1 similar comment
|
Rerun tests. |
rapids-bot bot
pushed a commit
to rapidsai/rmm
that referenced
this pull request
Jun 2, 2021
Removes default stream arguments from `rmm::device_buffer`. Now copy construction requires a stream argument (default copy ctor deleted), and copy assignment is disallowed (operator deleted). Move construction and assignment are still supported, and move assignment still use the most recently used stream for deallocating any previous data. Also improves device_buffer tests (implements TODOs in code). I don't think this should be merged until RAPIDS dependent libraries are ready for it. I have a libcudf PR in progress for this. Fixes #418 - [x] cuDF PR: rapidsai/cudf#8280 - [x] cuGraph PR: rapidsai/cugraph#1609 - [x] cuSpatial PR: rapidsai/cuspatial#403 - [x] cuML does not yet use device_buffer Authors: - Mark Harris (https://github.com/harrism) Approvers: - Rong Ou (https://github.com/rongou) - Keith Kraus (https://github.com/kkraus14) - Conor Hoekstra (https://github.com/codereport) - GALI PREM SAGAR (https://github.com/galipremsagar) URL: #775
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
In preparation for rapidsai/rmm#775, this PR has many changes to ensure streams on which device_buffers are allocated and copied are explicit. As a consequence, this PR also finds and fixes many places where buffers were copied that should have been moved. In addition, it adds copy constructors that take an optional stream to all
cudf::scalarclasses.This can and should be merged before rapidsai/rmm#775 to ensure no breakage of the build by that PR.