Github user JoshRosen commented on a diff in the pull request:
https://github.com/apache/spark/pull/8829#discussion_r40033212
--- Diff:
core/src/main/scala/org/apache/spark/shuffle/sort/SortShuffleManager.scala ---
@@ -19,10 +19,97 @@ package org.apache.spark.shuffle.sort
import java.util.concurrent.ConcurrentHashMap
-import org.apache.spark.{Logging, SparkConf, TaskContext,
ShuffleDependency}
+import org.apache.spark._
+import org.apache.spark.serializer.Serializer
import org.apache.spark.shuffle._
-import org.apache.spark.shuffle.hash.HashShuffleReader
+/**
+ * Subclass of [[BaseShuffleHandle]], used to identify when we've chosen
to use the new shuffle.
+ */
+private[spark] class UnsafeShuffleHandle[K, V](
+ shuffleId: Int,
+ numMaps: Int,
+ dependency: ShuffleDependency[K, V, V])
+ extends BaseShuffleHandle(shuffleId, numMaps, dependency) {
+}
+
+private[spark] object SortShuffleManager extends Logging {
+
+ /**
+ * The maximum number of shuffle output partitions that
UnsafeShuffleManager supports.
+ */
+ val MAX_SHUFFLE_OUTPUT_PARTITIONS =
PackedRecordPointer.MAXIMUM_PARTITION_ID + 1
+
+ /**
+ * Helper method for determining whether a shuffle should use the
optimized unsafe shuffle
+ * path or whether it should fall back to the original sort-based
shuffle.
+ */
+ def canUseUnsafeShuffle[K, V, C](dependency: ShuffleDependency[K, V,
C]): Boolean = {
+ val shufId = dependency.shuffleId
+ val serializer = Serializer.getSerializer(dependency.serializer)
+ if (!serializer.supportsRelocationOfSerializedObjects) {
+ log.debug(s"Can't use UnsafeShuffle for shuffle $shufId because the
serializer, " +
+ s"${serializer.getClass.getName}, does not support object
relocation")
+ false
+ } else if (dependency.aggregator.isDefined) {
+ log.debug(s"Can't use UnsafeShuffle for shuffle $shufId because an
aggregator is defined")
+ false
+ } else if (dependency.partitioner.numPartitions >
MAX_SHUFFLE_OUTPUT_PARTITIONS) {
+ log.debug(s"Can't use UnsafeShuffle for shuffle $shufId because it
has more than " +
+ s"$MAX_SHUFFLE_OUTPUT_PARTITIONS partitions")
+ false
+ } else {
+ log.debug(s"Can use UnsafeShuffle for shuffle $shufId")
+ true
+ }
+ }
+}
+
+/**
+ * A shuffle implementation that uses directly-managed memory to implement
several performance
--- End diff --
TODO: update these comments.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
