feat: Add native shuffle and columnar shuffle #30
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
|
The CI failure is: It will be fixed at #32 . |
viirya
force-pushed
the
native_shuffle
branch
from
b5956ae to
43ddbdb
Compare
|
cc @sunchao |
sunchao
reviewed
Feb 16, 2024
spark/src/main/scala/org/apache/spark/sql/comet/CometTakeOrderedAndProjectExec.scala
Outdated
Show resolved
Hide resolved
spark/src/main/scala/org/apache/comet/CometSparkSessionExtensions.scala
Outdated
Show resolved
Hide resolved
spark/src/main/scala/org/apache/comet/CometSparkSessionExtensions.scala
Outdated
Show resolved
Hide resolved
|
Let's also put some details in the PR description. |
|
Added some description there. |
|
Merged. Thanks. |
himadripal
pushed a commit
to himadripal/datafusion-comet
that referenced
this pull request
Sep 7, 2024
* build: Fix references to old Boson build * build: Only publish Comet for Spark 3.4
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Outdated suggestions cannot be applied.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.
Which issue does this PR close?
Closes #29.
Rationale for this change
As a columnar execution engine plugin to Spark, Comet comes out columnar operation support by providing corresponding operators to replace row-based operators in Spark. Shuffle is also a row-based operation in Spark. It is happened between the boundary of SQL operators which need to exchange data according to some specified distribution requirements. Without columnar shuffle, it means we need to do columnar to row/row to columnar around each shuffle operations. Thus, we propose Comet shuffle operators in this patch.
What changes are included in this PR?
Two kind of shuffle operators are included in this patch: native shuffle and columnar shuffle. Both shuffle operators are columnar-based operations and use same native implementation to write shuffle data into disk. Native shuffle takes columnar batches output from Comet operators directly. Columnar shuffle takes row outputs from downstream operators which could be Spark operators or Comet operators wrapped by ColumnarToRow operator. These rows are converted into columnar batches in the native writer and written into disk.
How are these changes tested?