Github user viirya commented on a diff in the pull request: https://github.com/apache/spark/pull/19763#discussion_r152908363 --- Diff: core/src/main/scala/org/apache/spark/internal/config/package.scala --- @@ -485,4 +485,13 @@ package object config { "array in the sorter.") .intConf .createWithDefault(Integer.MAX_VALUE) + + private[spark] val SHUFFLE_MAP_OUTPUT_PARALLEL_AGGREGATION_THRESHOLD = + ConfigBuilder("spark.shuffle.mapOutput.parallelAggregationThreshold") + .internal() + .doc("Multi-thread is used when the number of mappers * shuffle partitions is greater than " + + "or equal to this threshold.") --- End diff -- From above `statuses.length.toLong * totalSizes.length / parallelAggThreshold + 1`, looks like we need to have at least two times of this threshold to enable this parallel aggregation?
--- --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org