Align GpuUnionExec with Spark 4.1's partitioner-aware union behavior [databricks] #14164
There are no files selected for viewing
This file contains hidden or 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
This file contains hidden or 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
This file contains hidden or 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
| Original file line number | Diff line number | Diff line change |
|---|---|---|
|
|
@@ -37,7 +37,7 @@ import org.apache.spark.internal.Logging | |
| import org.apache.spark.rdd.RDD | ||
| import org.apache.spark.sql.catalyst.InternalRow | ||
| import org.apache.spark.sql.catalyst.expressions._ | ||
| import org.apache.spark.sql.catalyst.plans.physical.{Partitioning, PartitioningCollection, RangePartitioning, SinglePartition, UnknownPartitioning} | ||
| import org.apache.spark.sql.catalyst.util.{ArrayData, MapData} | ||
| import org.apache.spark.sql.execution.{FilterExec, ProjectExec, SampleExec, SparkPlan} | ||
| import org.apache.spark.sql.rapids.{GpuCreateArray, GpuCreateMap, GpuCreateNamedStruct, GpuPartitionwiseSampledRDD, GpuPoissonSampler} | ||
|
|
@@ -231,7 +231,53 @@ trait GpuProjectExecLike extends ShimUnaryExecNode with GpuExec { | |
|
|
||
| override def outputOrdering: Seq[SortOrder] = child.outputOrdering | ||
|
|
||
| /** | ||
| * Compute output partitioning, handling PartitioningCollection from joins. | ||
| * | ||
| * This matches Spark's PartitioningPreservingUnaryExecNode behavior: | ||
| * 1. Flatten any PartitioningCollection into individual partitionings | ||
| * 2. Filter to only keep partitionings whose attributes are in the output | ||
| * 3. Remap attributes through aliases | ||
| * | ||
| * This is critical for Spark 4.1+ where UnionExec uses outputPartitioning | ||
| * to decide between partitioner-aware union vs concatenation. | ||
| */ | ||
| override def outputPartitioning: Partitioning = { | ||
| val attributeMap = child.output.zip(output).toMap | ||
| val outputSet = AttributeSet(output) | ||
|
|
||
| // Flatten a PartitioningCollection into individual partitionings | ||
| def flattenPartitioning(p: Partitioning): Seq[Partitioning] = p match { | ||
| case PartitioningCollection(childPartitionings) => | ||
| childPartitionings.flatMap(flattenPartitioning) | ||
| case other => Seq(other) | ||
| } | ||
|
|
||
| def remapPartitioning(p: Partitioning): Partitioning = p match { | ||
| case e: Expression => | ||
| e.transform { | ||
| case a: Attribute if attributeMap.contains(a) => attributeMap(a) | ||
| }.asInstanceOf[Partitioning] | ||
| case other => other | ||
| } | ||
|
|
||
| val partitionings = flattenPartitioning(child.outputPartitioning).flatMap { | ||
| case e: Expression => | ||
| // Only keep partitionings whose attributes are all in the output | ||
| val remapped = remapPartitioning(e.asInstanceOf[Partitioning]) | ||
| remapped match { | ||
| case re: Expression if re.references.subsetOf(outputSet) => Some(remapped) | ||
| case _ => None | ||
| } | ||
| case other => Some(other) | ||
| } | ||
|
|
||
| partitionings match { | ||
| case Seq() => UnknownPartitioning(child.outputPartitioning.numPartitions) | ||
| case Seq(single) => single | ||
| case multiple => PartitioningCollection(multiple) | ||
| } | ||
| } | ||
|
|
||
| override def doExecute(): RDD[InternalRow] = | ||
| throw new IllegalStateException(s"Row-based execution should not occur for $this") | ||
|
|
@@ -1609,6 +1655,9 @@ case class GpuUnionExec(children: Seq[SparkPlan]) extends ShimSparkPlan with Gpu | |
| } | ||
| } | ||
|
|
||
| override def outputPartitioning: Partitioning = | ||
| GpuUnionExecShim.getOutputPartitioning(children, output, conf) | ||
|
Collaborator
There was a problem hiding this comment. Hmm, can we take the
Collaborator
Author
There was a problem hiding this comment. Thanks @jihoonson for the review. Your review comments on the refactor makes sense. I will do it as a follow-on PR for 26.04 if that's okay. Filed an issue for these - #14229 |
||
|
|
||
| // The smallest of our children | ||
| override def outputBatching: CoalesceGoal = | ||
| children.map(GpuExec.outputBatching).reduce(CoalesceGoal.minProvided) | ||
|
|
@@ -1620,11 +1669,12 @@ case class GpuUnionExec(children: Seq[SparkPlan]) extends ShimSparkPlan with Gpu | |
| val numOutputRows = gpuLongMetric(NUM_OUTPUT_ROWS) | ||
| val numOutputBatches = gpuLongMetric(NUM_OUTPUT_BATCHES) | ||
|
|
||
| GpuUnionExecShim.unionColumnarRdds( | ||
| sparkContext, | ||
| children.map(_.executeColumnar()), | ||
| outputPartitioning, | ||
| numOutputRows, | ||
| numOutputBatches) | ||
| } | ||
|
Collaborator
There was a problem hiding this comment. Do we need to also update what we say out output partitioning is? |
||
| } | ||
|
|
||
|
|
||
This file contains hidden or 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
This file contains hidden or 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
This file contains hidden or 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
This file contains hidden or 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
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,87 @@ | ||
| /* | ||
| * Copyright (c) 2026, NVIDIA CORPORATION. | ||
| * | ||
| * Licensed under the Apache License, Version 2.0 (the "License"); | ||
| * you may not use this file except in compliance with the License. | ||
| * You may obtain a copy of the License at | ||
| * | ||
| * http://www.apache.org/licenses/LICENSE-2.0 | ||
| * | ||
| * Unless required by applicable law or agreed to in writing, software | ||
| * distributed under the License is distributed on an "AS IS" BASIS, | ||
| * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. | ||
| * See the License for the specific language governing permissions and | ||
| * limitations under the License. | ||
| */ | ||
|
|
||
| /*** spark-rapids-shim-json-lines | ||
| {"spark": "320"} | ||
| {"spark": "321"} | ||
| {"spark": "321cdh"} | ||
| {"spark": "322"} | ||
| {"spark": "323"} | ||
| {"spark": "324"} | ||
| {"spark": "330"} | ||
| {"spark": "330cdh"} | ||
| {"spark": "330db"} | ||
| {"spark": "331"} | ||
| {"spark": "332"} | ||
| {"spark": "332cdh"} | ||
| {"spark": "332db"} | ||
| {"spark": "333"} | ||
| {"spark": "334"} | ||
| {"spark": "340"} | ||
| {"spark": "341"} | ||
| {"spark": "341db"} | ||
| {"spark": "342"} | ||
| {"spark": "343"} | ||
| {"spark": "344"} | ||
| {"spark": "350"} | ||
| {"spark": "350db143"} | ||
| {"spark": "351"} | ||
| {"spark": "352"} | ||
| {"spark": "353"} | ||
| {"spark": "354"} | ||
| {"spark": "355"} | ||
| {"spark": "356"} | ||
| {"spark": "357"} | ||
| {"spark": "400"} | ||
| {"spark": "401"} | ||
| spark-rapids-shim-json-lines ***/ | ||
| package com.nvidia.spark.rapids.shims | ||
|
Collaborator
There was a problem hiding this comment. missing {"spark": "400db173"} ?
Collaborator
Author
There was a problem hiding this comment. Added [databricks] marker in the PR title. I will add "400db173" in the PR to support for Databricks-17.3. Adding now itself might be confusing for some. |
||
|
|
||
| import com.nvidia.spark.rapids.GpuMetric | ||
|
|
||
| import org.apache.spark.SparkContext | ||
| import org.apache.spark.rdd.RDD | ||
| import org.apache.spark.sql.catalyst.expressions.Attribute | ||
| import org.apache.spark.sql.catalyst.plans.physical.{Partitioning, UnknownPartitioning} | ||
| import org.apache.spark.sql.execution.SparkPlan | ||
| import org.apache.spark.sql.internal.SQLConf | ||
| import org.apache.spark.sql.vectorized.ColumnarBatch | ||
|
|
||
| /** | ||
| * Pre-Spark 4.1 implementation: uses sparkContext.union() which concatenates partitions. | ||
| */ | ||
| object GpuUnionExecShim { | ||
|
|
||
| def unionColumnarRdds( | ||
| sc: SparkContext, | ||
| rdds: Seq[RDD[ColumnarBatch]], | ||
| outputPartitioning: Partitioning, | ||
| numOutputRows: GpuMetric, | ||
| numOutputBatches: GpuMetric): RDD[ColumnarBatch] = { | ||
| sc.union(rdds).map { batch => | ||
| numOutputBatches += 1 | ||
| numOutputRows += batch.numRows | ||
| batch | ||
| } | ||
| } | ||
|
|
||
| def getOutputPartitioning( | ||
| children: Seq[SparkPlan], | ||
| unionOutput: Seq[Attribute], | ||
| conf: SQLConf): Partitioning = { | ||
| UnknownPartitioning(0) | ||
| } | ||
| } | ||
Oops, something went wrong.
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.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Do we not need to handle the case when the project has aliases?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Also wonder if it's a good idea to have
GpuProjectExecLikeextendPartitioningPreservingUnaryExecNodeinstead of copying this code.