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

Add retry support in sub hash join #11706

Merged
merged 1 commit into from
Nov 7, 2024
Merged

Add retry support in sub hash join #11706

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
16 changes: 13 additions & 3 deletions ...plugin/src/main/scala/org/apache/spark/sql/rapids/execution/GpuSubPartitionHashJoin.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -18,7 +18,7 @@ package org.apache.spark.sql.rapids.execution
import scala.collection.mutable
import scala.collection.mutable.ArrayBuffer

import com.nvidia.spark.rapids.{GpuBatchUtils, GpuColumnVector, GpuExpression, GpuHashPartitioningBase, GpuMetric, SpillableColumnarBatch, SpillPriorities, TaskAutoCloseableResource}
import com.nvidia.spark.rapids.{GpuBatchUtils, GpuColumnVector, GpuExpression, GpuHashPartitioningBase, GpuMetric, RmmRapidsRetryIterator, SpillableColumnarBatch, SpillPriorities, TaskAutoCloseableResource}
import com.nvidia.spark.rapids.Arm.{closeOnExcept, withResource}
import com.nvidia.spark.rapids.RapidsPluginImplicits._

Expand Down Expand Up @@ -179,9 +179,19 @@ class GpuBatchSubPartitioner(
// 1) Hash partition on the batch
val partedTable = GpuHashPartitioningBase.hashPartitionAndClose(
gpuBatch, inputBoundKeys, realNumPartitions, "Sub-Hash Calculate", hashSeed)
val (spillBatch, partitions) = withResource(partedTable) { _ =>
// Convert to SpillableColumnarBatch for the following retry.
(SpillableColumnarBatch(GpuColumnVector.from(partedTable.getTable, types),
SpillPriorities.ACTIVE_BATCHING_PRIORITY),
partedTable.getPartitions)
}
// 2) Split into smaller tables according to partitions
val subTables = withResource(partedTable) { _ =>
partedTable.getTable.contiguousSplit(partedTable.getPartitions.tail: _*)
val subTables = RmmRapidsRetryIterator.withRetryNoSplit(spillBatch) { _ =>
withResource(spillBatch.getColumnarBatch()) { cb =>
withResource(GpuColumnVector.from(cb)) { tbl =>
tbl.contiguousSplit(partitions.tail: _*)
}
}
}
// 3) Make each smaller table spillable and cache them in the queue
withResource(subTables) { _ =>
Expand Down
Loading