marin-ma commented on code in PR #12087:
URL: https://github.com/apache/gluten/pull/12087#discussion_r3316376469
##########
backends-velox/src/main/scala/org/apache/gluten/extension/AppendBatchResizeForShuffleInputAndOutput.scala:
##########
@@ -22,17 +22,18 @@ import org.apache.gluten.execution.VeloxResizeBatchesExec
import org.apache.spark.sql.catalyst.rules.Rule
import org.apache.spark.sql.execution.{ColumnarShuffleExchangeExec, SparkPlan}
import org.apache.spark.sql.execution.adaptive.{AQEShuffleReadExec,
ShuffleQueryStageExec}
-import org.apache.spark.sql.execution.exchange.ReusedExchangeExec
/**
* Try to append [[VeloxResizeBatchesExec]] for shuffle input and output to
make the batch sizes in
* good shape.
*/
-case class AppendBatchResizeForShuffleInputAndOutput() extends Rule[SparkPlan]
{
+case class AppendBatchResizeForShuffleInputAndOutput(isAdaptiveContext:
Boolean)
Review Comment:
In the original logic, the "resize batches for shuffle output" only applies
to aqe stage nodes, so `addResizeBatchesForShuffleOutput` is a no-op when AQE
is off.
Thanks for catching and this should be added
[here](https://github.com/apache/gluten/pull/12087/changes#diff-94c14147622eeb96705fe5fc473a0d6876edb5d18cb4f412e168134f44feec6cR52)
to indicate the rule doesn't apply when AQE is off.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
