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

Remove InMemoryTableScanExec support for Spark 3.5+ [databricks] #10602

Merged
merged 4 commits into from
Mar 18, 2024
Merged

Remove InMemoryTableScanExec support for Spark 3.5+ [databricks] #10602

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
8d38f8a
Remove InMemoryTableScanExec support for Spark 3.5+
razajafri Mar 14, 2024
c230679
Signing off
razajafri Mar 15, 2024
4f504e3
Revert "Remove InMemoryTableScanExec support for Spark 3.5+"
razajafri Mar 15, 2024
41fc95a
Disable InMemoryTableScanExec by default for Spark versions 3.5+
razajafri Mar 15, 2024
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
17 changes: 15 additions & 2 deletions sql-plugin/src/main/spark350/scala/com/nvidia/spark/rapids/shims/SparkShims.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -24,6 +24,7 @@ import com.nvidia.spark.rapids._

import org.apache.spark.sql.catalyst.expressions.{Expression, PythonUDAF, ToPrettyString}
import org.apache.spark.sql.execution.SparkPlan
import org.apache.spark.sql.execution.columnar.InMemoryTableScanExec
import org.apache.spark.sql.execution.window.WindowGroupLimitExec
import org.apache.spark.sql.rapids.execution.python.GpuPythonUDAF
import org.apache.spark.sql.types.StringType
Expand Down Expand Up @@ -78,13 +79,25 @@ object SparkShimImpl extends Spark340PlusNonDBShims {
}

override def getExecs: Map[Class[_ <: SparkPlan], ExecRule[_ <: SparkPlan]] = {
val shimExecs: Map[Class[_ <: SparkPlan], ExecRule[_ <: SparkPlan]]= Seq(
val imtsKey = classOf[InMemoryTableScanExec].asSubclass(classOf[SparkPlan])
// To avoid code duplication we are reusing the rule from GpuOverrides
// but we disable it by default
val imtsRule = GpuOverrides.commonExecs.getOrElse(imtsKey,
throw new IllegalStateException("InMemoryTableScan should be overridden by default before" +
" Spark 3.5.0")).
disabledByDefault(
"""there could be complications when using it with AQE with Spark-3.5.0 and Spark-3.5.1.
|For more details please check
|https://github.com/NVIDIA/spark-rapids/issues/10603""".stripMargin.replaceAll("\n", " "))

val shimExecs: Map[Class[_ <: SparkPlan], ExecRule[_ <: SparkPlan]] = Seq(
imtsRule,
GpuOverrides.exec[WindowGroupLimitExec](
"Apply group-limits for row groups destined for rank-based window functions like " +
"row_number(), rank(), and dense_rank()",
ExecChecks( // Similar to WindowExec.
(TypeSig.commonCudfTypes + TypeSig.NULL + TypeSig.DECIMAL_128 +
TypeSig.STRUCT + TypeSig.ARRAY + TypeSig.MAP).nested(),
TypeSig.STRUCT + TypeSig.ARRAY + TypeSig.MAP).nested(),
TypeSig.all),
(limit, conf, p, r) => new GpuWindowGroupLimitExecMeta(limit, conf, p, r))
).map(r => (r.getClassFor.asSubclass(classOf[SparkPlan]), r)).toMap
Expand Down
2 changes: 1 addition & 1 deletion tools/generated_files/350/supportedExecs.csv
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -18,7 +18,7 @@ AQEShuffleReadExec,S,None,Input/Output,S,S,S,S,S,S,S,S,PS,S,S,S,S,NS,PS,PS,PS,NS
HashAggregateExec,S,None,Input/Output,S,S,S,S,S,S,S,S,PS,S,S,S,PS,NS,PS,PS,PS,NS,NS,NS
ObjectHashAggregateExec,S,None,Input/Output,S,S,S,S,S,S,S,S,PS,S,S,S,PS,NS,PS,PS,PS,NS,NS,NS
SortAggregateExec,S,None,Input/Output,S,S,S,S,S,S,S,S,PS,S,S,S,PS,NS,PS,PS,PS,NS,NS,NS
InMemoryTableScanExec,S,None,Input/Output,S,S,S,S,S,S,S,S,PS,S,S,NS,NS,NS,PS,PS,PS,NS,S,S
InMemoryTableScanExec,NS,This is disabled by default because there could be complications when using it with AQE. For more details please check https://github.com/NVIDIA/spark-rapids/issues/10603,Input/Output,S,S,S,S,S,S,S,S,PS,S,S,NS,NS,NS,PS,PS,PS,NS,S,S
DataWritingCommandExec,S,None,Input/Output,S,S,S,S,S,S,S,S,PS,S,PS,NS,S,NS,PS,PS,PS,NS,S,S
ExecutedCommandExec,S,None,Input/Output,S,S,S,S,S,S,S,S,PS,S,S,S,S,S,PS,PS,PS,S,S,S
WriteFilesExec,S,None,Input/Output,S,S,S,S,S,S,S,S,PS,S,S,S,S,S,PS,PS,PS,S,S,S
Expand Down
Loading