ulysses-you commented on code in PR #57742:
URL: https://github.com/apache/spark/pull/57742#discussion_r3712197261
##########
sql/core/src/main/scala/org/apache/spark/sql/execution/aggregate/HashAggregateExec.scala:
##########
@@ -663,46 +855,122 @@ case class HashAggregateExec(
case _ => ("true", "", "")
}
- val findOrInsertRegularHashMap: String =
- s"""
- |// generate grouping key
- |${unsafeRowKeyCode.code}
- |int $unsafeRowKeyHash = ${unsafeRowKeyCode.value}.hashCode();
- |if ($checkFallbackForBytesToBytesMap) {
- | // try to get the buffer from hash map
- | $unsafeRowBuffer =
- | $hashMapTerm.getAggregationBufferFromUnsafeRow($unsafeRowKeys,
$unsafeRowKeyHash);
- |}
- |// Can't allocate buffer from the hash map. Spill the map and
fallback to sort-based
- |// aggregation after processing all input rows.
- |if ($unsafeRowBuffer == null) {
- | if ($sorterTerm == null) {
- | $sorterTerm = $hashMapTerm.destructAndCreateExternalSorter();
- | } else {
- |
$sorterTerm.merge($hashMapTerm.destructAndCreateExternalSorter());
- | }
- | $resetCounter
- | // the hash map had be spilled, it should have enough memory now,
- | // try to allocate buffer again.
- | $unsafeRowBuffer = $hashMapTerm.getAggregationBufferFromUnsafeRow(
- | $unsafeRowKeys, $unsafeRowKeyHash);
- | if ($unsafeRowBuffer == null) {
- | // failed to allocate the first page
- | throw QueryExecutionErrors.aggregateOutOfMemoryError();
- | }
- |}
- """.stripMargin
+ val findOrInsertRegularHashMap: String = {
+ // Assumes the grouping key projection (`unsafeRowKeyCode.code`) has
already run for this row,
+ // so `unsafeRowKeyCode.value` holds the current key. The projection is
emitted exactly once
+ // per regular-map row (see below); emitting it in more than one runtime
branch is unsafe
+ // because the projection's subexpression/writer state assigned in one
branch would be read
+ // stale from another (e.g. the adaptive pass-through path would reuse
the last probed key).
+ val probeRegularMap =
+ s"""
+ |int $unsafeRowKeyHash = ${unsafeRowKeyCode.value}.hashCode();
+ |if ($checkFallbackForBytesToBytesMap) {
+ | // try to get the buffer from hash map
+ | $unsafeRowBuffer =
+ | $hashMapTerm.getAggregationBufferFromUnsafeRow($unsafeRowKeys,
$unsafeRowKeyHash);
+ |}
+ """.stripMargin
+
+ val spillMap =
+ s"""
+ |if ($sorterTerm == null) {
+ | $sorterTerm = $hashMapTerm.destructAndCreateExternalSorter();
+ |} else {
+ |
$sorterTerm.merge($hashMapTerm.destructAndCreateExternalSorter());
+ |}
+ |$resetCounter
+ |// the hash map had be spilled, it should have enough memory now,
Review Comment:
Addressed
##########
sql/core/src/test/scala/org/apache/spark/sql/execution/benchmark/AdaptivePartialAggregationBenchmark.scala:
##########
@@ -0,0 +1,137 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements. See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.apache.spark.sql.execution.benchmark
+
+import org.apache.spark.benchmark.Benchmark
+import org.apache.spark.sql.DataFrame
+import org.apache.spark.sql.internal.SQLConf
+
+/**
+ * Benchmark comparing runtime adaptive partial aggregation (see
+ * [[SQLConf.ADAPTIVE_PARTIAL_AGGREGATION_ENABLED]]) against the static
pre-shuffle partial
+ * aggregation. When the partial aggregation is not reducing rows, the
operator streams the
+ * remaining rows through as single-row partial buffers instead of maintaining
(and possibly
+ * spilling) a large aggregation map.
+ *
+ * Each scenario runs the query across the full matrix of whole-stage codegen
on/off and the
+ * feature disabled (`adaptive = F`, the pre-change baseline) vs enabled
(`adaptive = T`), over a
+ * {high, low}-cardinality x {no-spill, on-spill} grid:
+ * - high-cardinality, no spill: the no-spill tier bypasses, which should
win.
+ * - low-cardinality, no spill: nothing bypasses, which must not regress.
+ * - high-cardinality, forced regular-map spill: the on-spill tier bypasses
instead of spilling,
+ * which should win.
+ * - low-cardinality, forced regular-map spill: the ratio is too low for the
on-spill tier to
+ * bypass, so both runs spill identically (no regression).
+ *
+ * To run this benchmark:
+ * {{{
+ * 1. build/sbt "sql/Test/runMain
+ *
org.apache.spark.sql.execution.benchmark.AdaptivePartialAggregationBenchmark"
+ * 2. generate result: SPARK_GENERATE_BENCHMARK_FILES=1 build/sbt
"sql/Test/runMain
+ *
org.apache.spark.sql.execution.benchmark.AdaptivePartialAggregationBenchmark"
+ * Results will be written to
"benchmarks/AdaptivePartialAggregationBenchmark-results.txt".
+ * }}}
+ */
+object AdaptivePartialAggregationBenchmark extends SqlBasedBenchmark {
+
+ override def runBenchmarkSuite(mainArgs: Array[String]): Unit = {
+ // The upstream `CombineAdjacentAggregation` and `ReplaceHashWithSortAgg`
rules would collapse
+ // or convert these single-partition hash aggregates, so both are disabled
to keep the
+ // Partial+Final `HashAggregateExec` structure the adaptive feature
governs.
+ val fixedPlanConfs = Seq(
+ SQLConf.COMBINE_ADJACENT_AGGREGATION_ENABLED.key -> "false",
+ SQLConf.REPLACE_HASH_WITH_SORT_AGG_ENABLED.key -> "false")
+
+ // Adds the (whole-stage codegen, adaptive switch) matrix for `query`.
`extraConf` is applied
+ // to all four cases so the only differences are the two axes.
+ def addCodegenAdaptiveCases(
+ benchmark: Benchmark,
+ query: () => DataFrame,
+ extraConf: Seq[(String, String)] = Nil): Unit = {
+ for {
+ wholeStage <- Seq(true, false)
+ adaptive <- Seq(false, true)
+ } {
+ val adaptiveLabel = if (adaptive) "T" else "F"
+ val label = s"codegen = $wholeStage, adaptive = $adaptiveLabel"
+ benchmark.addCase(label) { _ =>
+ withSQLConf(
+ (Seq(
+ SQLConf.WHOLESTAGE_CODEGEN_ENABLED.key -> wholeStage.toString,
+ SQLConf.ADAPTIVE_PARTIAL_AGGREGATION_ENABLED.key ->
adaptive.toString) ++
+ fixedPlanConfs ++ extraConf): _*) {
+ query().noop()
+ }
+ }
+ }
+ }
+
+ // Fully distinct keys make partial aggregation useless, so the no-spill
(Tier 1) sampling tier
+ // bypasses: the feature should be faster than the baseline that maintains
a map entry per row.
+ runBenchmark("high-cardinality input, no-spill pass-through (Tier 1)") {
+ val N = 8L << 20
+ val benchmark = new Benchmark("adaptive partial agg, high card, no
spill", N,
+ output = output)
+ addCodegenAdaptiveCases(benchmark, () => distinctKeyedDf(N))
+ benchmark.run()
+ }
+
+ // 1000 distinct keys over a large input: partial aggregation reduces a
lot, the no-spill tier
+ // never fires, and the two runs must match (no regression).
+ runBenchmark("low-cardinality input, no-spill pass-through (Tier 1)") {
+ val N = 16L << 20
+ val benchmark = new Benchmark("adaptive partial agg, low card, no
spill", N,
+ output = output)
+ addCodegenAdaptiveCases(benchmark, () =>
+ spark.range(N).selectExpr("id % 1000 as k", "id as
v").groupBy("k").agg("v" -> "sum"))
+ benchmark.run()
+ }
+
+ // Force the regular map to spill quickly and disable the no-spill tier
(huge sample). With
+ // fully distinct keys the reduction ratio is 1.0, so at the spill
boundary the on-spill
+ // (Tier 2) tier bypasses instead of spilling; the baseline spills
repeatedly and falls back
Review Comment:
Addressed
--
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]
