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

[SPARK-33523][SQL][TEST] Add predicate related benchmark to SubExprEliminationBenchmark #30476

Closed
wants to merge 3 commits into from
Closed

[SPARK-33523][SQL][TEST] Add predicate related benchmark to SubExprEliminationBenchmark #30476

Show file tree
Hide file tree
Changes from 2 commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
7e3be61
Add predicate benchmark.
viirya Nov 22, 2020
0ca10ba
Use noop source.
viirya Nov 24, 2020
930ffe2
Refactor.
viirya Nov 24, 2020
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
20 changes: 15 additions & 5 deletions sql/core/benchmarks/SubExprEliminationBenchmark-jdk11-results.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -5,11 +5,21 @@ Benchmark for performance of subexpression elimination
Preparing data for benchmarking ...
OpenJDK 64-Bit Server VM 11.0.9+11 on Mac OS X 10.15.6
Intel(R) Core(TM) i7-9750H CPU @ 2.60GHz
from_json as subExpr: Best Time(ms) Avg Time(ms) Stdev(ms) Rate(M/s) Per Row(ns) Relative
from_json as subExpr in Project: Best Time(ms) Avg Time(ms) Stdev(ms) Rate(M/s) Per Row(ns) Relative
-------------------------------------------------------------------------------------------------------------------------
subexpressionElimination off, codegen on 25932 26908 916 0.0 259320042.3 1.0X
subexpressionElimination off, codegen off 26085 26159 65 0.0 260848905.0 1.0X
subexpressionElimination on, codegen on 2860 2939 72 0.0 28603312.9 9.1X
subexpressionElimination on, codegen off 2517 2617 93 0.0 25165157.7 10.3X
subexpressionElimination off, codegen on 22605 22935 291 0.0 226047196.5 1.0X
subexpressionElimination off, codegen off 21811 22151 303 0.0 218105716.6 1.0X
subexpressionElimination on, codegen on 1353 1385 36 0.0 13531011.3 16.7X
subexpressionElimination on, codegen off 1237 1260 20 0.0 12368657.3 18.3X

Preparing data for benchmarking ...
OpenJDK 64-Bit Server VM 11.0.9+11 on Mac OS X 10.15.6
Intel(R) Core(TM) i7-9750H CPU @ 2.60GHz
from_json as subExpr in Filter: Best Time(ms) Avg Time(ms) Stdev(ms) Rate(M/s) Per Row(ns) Relative
-------------------------------------------------------------------------------------------------------------------------
subexpressionElimination off, codegen on 32792 33101 282 0.0 327922763.5 1.0X
subexpressionElimination off, codegen off 32809 33433 550 0.0 328088662.6 1.0X
subexpressionElimination on, codegen on 18173 18828 869 0.0 181734709.5 1.8X
subexpressionElimination on, codegen off 33695 33951 287 0.0 336950807.7 1.0X


20 changes: 15 additions & 5 deletions sql/core/benchmarks/SubExprEliminationBenchmark-results.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -5,11 +5,21 @@ Benchmark for performance of subexpression elimination
Preparing data for benchmarking ...
OpenJDK 64-Bit Server VM 1.8.0_265-b01 on Mac OS X 10.15.6
Intel(R) Core(TM) i7-9750H CPU @ 2.60GHz
from_json as subExpr: Best Time(ms) Avg Time(ms) Stdev(ms) Rate(M/s) Per Row(ns) Relative
from_json as subExpr in Project: Best Time(ms) Avg Time(ms) Stdev(ms) Rate(M/s) Per Row(ns) Relative
-------------------------------------------------------------------------------------------------------------------------
subexpressionElimination off, codegen on 26503 27622 1937 0.0 265033362.4 1.0X
subexpressionElimination off, codegen off 24920 25376 430 0.0 249196978.2 1.1X
subexpressionElimination on, codegen on 2421 2466 39 0.0 24213606.1 10.9X
subexpressionElimination on, codegen off 2360 2435 87 0.0 23604320.7 11.2X
subexpressionElimination off, codegen on 25887 26105 326 0.0 258868246.6 1.0X
subexpressionElimination off, codegen off 25131 25454 522 0.0 251309329.7 1.0X
subexpressionElimination on, codegen on 2230 2340 106 0.0 22302959.3 11.6X
subexpressionElimination on, codegen off 2185 2254 64 0.0 21852694.0 11.8X

Preparing data for benchmarking ...
OpenJDK 64-Bit Server VM 1.8.0_265-b01 on Mac OS X 10.15.6
Intel(R) Core(TM) i7-9750H CPU @ 2.60GHz
from_json as subExpr in Filter: Best Time(ms) Avg Time(ms) Stdev(ms) Rate(M/s) Per Row(ns) Relative
-------------------------------------------------------------------------------------------------------------------------
subexpressionElimination off, codegen on 42687 42805 111 0.0 426873372.5 1.0X
subexpressionElimination off, codegen off 43606 45108 1613 0.0 436055236.3 1.0X
subexpressionElimination on, codegen on 29761 30563 704 0.0 297614324.4 1.4X
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

In this case, do we have only 40% improvement?

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Seems so, it is subexpression elimination under codegen.

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I will submit subexpression elimination under non-codegen after this.

subexpressionElimination on, codegen off 41676 42598 955 0.0 416758112.3 1.0X


82 changes: 77 additions & 5 deletions sql/core/src/test/scala/org/apache/spark/sql/execution/SubExprEliminationBenchmark.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -17,6 +17,8 @@
package org.apache.spark.sql.execution

import org.apache.spark.benchmark.Benchmark
import org.apache.spark.sql.Column
import org.apache.spark.sql.catalyst.expressions.{Expression, Literal, Or}
import org.apache.spark.sql.execution.benchmark.SqlBasedBenchmark
import org.apache.spark.sql.functions._
import org.apache.spark.sql.internal.SQLConf
Expand All @@ -39,7 +41,7 @@ object SubExprEliminationBenchmark extends SqlBasedBenchmark {
import spark.implicits._

def withFromJson(rowsNum: Int, numIters: Int): Unit = {
val benchmark = new Benchmark("from_json as subExpr", rowsNum, output = output)
val benchmark = new Benchmark("from_json as subExpr in Project", rowsNum, output = output)

withTempPath { path =>
prepareDataInfo(benchmark)
Expand All @@ -61,7 +63,7 @@ object SubExprEliminationBenchmark extends SqlBasedBenchmark {
val df = spark.read
.text(path.getAbsolutePath)
.select(cols: _*)
df.collect()
df.write.mode("overwrite").format("noop").save()
}
}

Expand All @@ -74,7 +76,7 @@ object SubExprEliminationBenchmark extends SqlBasedBenchmark {
val df = spark.read
.text(path.getAbsolutePath)
.select(cols: _*)
df.collect()
df.write.mode("overwrite").format("noop").save()
}
}

Expand All @@ -87,7 +89,7 @@ object SubExprEliminationBenchmark extends SqlBasedBenchmark {
val df = spark.read
.text(path.getAbsolutePath)
.select(cols: _*)
df.collect()
df.write.mode("overwrite").format("noop").save()
}
}

Expand All @@ -100,19 +102,89 @@ object SubExprEliminationBenchmark extends SqlBasedBenchmark {
val df = spark.read
.text(path.getAbsolutePath)
.select(cols: _*)
df.collect()
df.write.mode("overwrite").format("noop").save()
}
}

benchmark.run()
}
}

def withFilter(rowsNum: Int, numIters: Int): Unit = {
val benchmark = new Benchmark("from_json as subExpr in Filter", rowsNum, output = output)

withTempPath { path =>
prepareDataInfo(benchmark)
val numCols = 1000
val schema = writeWideRow(path.getAbsolutePath, rowsNum, numCols)

val predicate = (0 until numCols).map { idx =>
(from_json('value, schema).getField(s"col$idx") >= Literal(100000)).expr
}.asInstanceOf[Seq[Expression]].reduce(Or)

// We only benchmark subexpression performance under codegen/non-codegen, so disabling
// json optimization.
benchmark.addCase("subexpressionElimination off, codegen on", numIters) { _ =>
withSQLConf(
SQLConf.SUBEXPRESSION_ELIMINATION_ENABLED.key -> "false",
HyukjinKwon marked this conversation as resolved.
Show resolved Hide resolved
SQLConf.WHOLESTAGE_CODEGEN_ENABLED.key -> "true",
SQLConf.CODEGEN_FACTORY_MODE.key -> "CODEGEN_ONLY",
SQLConf.JSON_EXPRESSION_OPTIMIZATION.key -> "false") {
val df = spark.read
.text(path.getAbsolutePath)
.where(Column(predicate))
df.write.mode("overwrite").format("noop").save()
}
}

benchmark.addCase("subexpressionElimination off, codegen off", numIters) { _ =>
withSQLConf(
SQLConf.SUBEXPRESSION_ELIMINATION_ENABLED.key -> "false",
SQLConf.WHOLESTAGE_CODEGEN_ENABLED.key -> "false",
SQLConf.CODEGEN_FACTORY_MODE.key -> "NO_CODEGEN",
SQLConf.JSON_EXPRESSION_OPTIMIZATION.key -> "false") {
val df = spark.read
.text(path.getAbsolutePath)
.where(Column(predicate))
df.write.mode("overwrite").format("noop").save()
}
}

benchmark.addCase("subexpressionElimination on, codegen on", numIters) { _ =>
withSQLConf(
SQLConf.SUBEXPRESSION_ELIMINATION_ENABLED.key -> "true",
SQLConf.WHOLESTAGE_CODEGEN_ENABLED.key -> "true",
SQLConf.CODEGEN_FACTORY_MODE.key -> "CODEGEN_ONLY",
SQLConf.JSON_EXPRESSION_OPTIMIZATION.key -> "false") {
val df = spark.read
.text(path.getAbsolutePath)
.where(Column(predicate))
df.write.mode("overwrite").format("noop").save()
}
}

benchmark.addCase("subexpressionElimination on, codegen off", numIters) { _ =>
withSQLConf(
SQLConf.SUBEXPRESSION_ELIMINATION_ENABLED.key -> "true",
SQLConf.WHOLESTAGE_CODEGEN_ENABLED.key -> "false",
SQLConf.CODEGEN_FACTORY_MODE.key -> "NO_CODEGEN",
SQLConf.JSON_EXPRESSION_OPTIMIZATION.key -> "false") {
val df = spark.read
.text(path.getAbsolutePath)
.where(Column(predicate))
df.write.mode("overwrite").format("noop").save()
}
}

benchmark.run()
}
}

override def runBenchmarkSuite(mainArgs: Array[String]): Unit = {
val numIters = 3
runBenchmark("Benchmark for performance of subexpression elimination") {
withFromJson(100, numIters)
withFilter(100, numIters)
}
}
}