Github user marmbrus commented on a diff in the pull request:
https://github.com/apache/spark/pull/7940#discussion_r36248632
--- Diff:
sql/core/src/main/scala/org/apache/spark/sql/execution/joins/BroadcastHashJoin.scala
---
@@ -59,19 +57,17 @@ case class BroadcastHashJoin(
override def requiredChildDistribution: Seq[Distribution] =
UnspecifiedDistribution :: UnspecifiedDistribution :: Nil
- @transient
- private val broadcastFuture = future {
+ private lazy val broadcastedRelation = {
--- End diff --
There was a noticeable speed-up in TPC-DS from broadcasting in parallel and
I think this removes that. Perhaps we should both use a `lazy val` and a
`future`. We can have the prepare phase make sure the future all get invoked
before running.
---
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]
