Copilot commented on code in PR #13115: URL: https://github.com/apache/gluten/pull/13115#discussion_r4278528950
########## gluten-ut/spark35/src/test/backends-velox/org/apache/spark/sql/execution/joins/GlutenParallelBroadcastJoinSuite.scala: ########## @@ -0,0 +1,65 @@ +/* + * 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.joins + +import org.apache.gluten.execution.BroadcastHashJoinExecTransformer + +import org.apache.spark.SparkConf +import org.apache.spark.sql.GlutenSQLTestsTrait +import org.apache.spark.sql.execution.ColumnarBroadcastExchangeExec + +class GlutenParallelBroadcastJoinSuite extends GlutenSQLTestsTrait { + + override def sparkConf: SparkConf = super.sparkConf + .set("spark.sql.adaptive.enabled", "false") + .set("spark.gluten.sql.columnar.backend.velox.IOThreads", "0") + .set("spark.gluten.velox.broadcast.build.targetBytesPerThread", "8KB") + + for (buildOnDriver <- Seq(false, true)) { + testGluten(s"parallel broadcast hash table build with IO disabled, driver=$buildOnDriver") { + withSQLConf( + "spark.gluten.sql.columnar.backend.velox.driverSideBroadcastHashTableBuild" -> + buildOnDriver.toString, + "spark.gluten.velox.minTableRowsForParallelJoinBuild" -> "1", + "spark.gluten.sql.columnar.backend.velox.broadcastHashTableBuildThreads" -> "2" Review Comment: The test config forces `broadcastHashTableBuildThreads` to `"1"` but later asserts `buildThreads > 1`, which makes the assertion inconsistent with the configured pool size. To make the test deterministic, either (a) set `broadcastHashTableBuildThreads` to a value > 1 (e.g., 2) and assert `>= 2` (or `== 2` if the metric is expected to match exactly), or (b) stop overriding the thread count in `withSQLConf` and assert based on the default sizing behavior. ########## cpp/velox/compute/VeloxBackend.cc: ########## @@ -254,6 +255,10 @@ void VeloxBackend::init( std::make_unique<folly::CPUThreadPoolExecutor>(ioThreads, folly::CPUThreadPoolExecutor::makeLifoSemQueue()); } + const auto broadcastHashTableBuildThreads = + std::max<int32_t>(1, backendConf_->get<int32_t>(kBroadcastHashTableBuildThreads, numTaskSlotsPerExecutor)); + broadcastHashTableBuildExecutor_ = std::make_unique<folly::CPUThreadPoolExecutor>(broadcastHashTableBuildThreads); Review Comment: Defaulting the dedicated CPU pool size to `numTaskSlotsPerExecutor` can create a large number of extra threads per executor (in addition to existing pools like IO/spill), which can cause oversubscription and higher memory/context-switch overhead when task slots are configured higher than available cores. Consider clamping the default to a CPU-based upper bound (e.g., hardware concurrency / configured executor cores) while still allowing users to opt into larger values via `kBroadcastHashTableBuildThreads`. ########## gluten-ut/spark35/src/test/backends-velox/org/apache/spark/sql/execution/joins/GlutenParallelBroadcastJoinSuite.scala: ########## @@ -0,0 +1,65 @@ +/* + * 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.joins + +import org.apache.gluten.execution.BroadcastHashJoinExecTransformer + +import org.apache.spark.SparkConf +import org.apache.spark.sql.GlutenSQLTestsTrait +import org.apache.spark.sql.execution.ColumnarBroadcastExchangeExec + +class GlutenParallelBroadcastJoinSuite extends GlutenSQLTestsTrait { + + override def sparkConf: SparkConf = super.sparkConf + .set("spark.sql.adaptive.enabled", "false") + .set("spark.gluten.sql.columnar.backend.velox.IOThreads", "0") + .set("spark.gluten.velox.broadcast.build.targetBytesPerThread", "8KB") + + for (buildOnDriver <- Seq(false, true)) { + testGluten(s"parallel broadcast hash table build with IO disabled, driver=$buildOnDriver") { + withSQLConf( + "spark.gluten.sql.columnar.backend.velox.driverSideBroadcastHashTableBuild" -> + buildOnDriver.toString, + "spark.gluten.velox.minTableRowsForParallelJoinBuild" -> "1", + "spark.gluten.sql.columnar.backend.velox.broadcastHashTableBuildThreads" -> "2" + ) { + withTable("parallel_broadcast_build") { + spark.range(4096) + .selectExpr("id", "sha2(cast(id as string), 256) AS value") + .write.saveAsTable("parallel_broadcast_build") + + val result = spark.sql(""" + |SELECT /*+ BROADCAST(b) */ p.id, b.value + |FROM parallel_broadcast_build p JOIN parallel_broadcast_build b + |ON p.id = b.id + |""".stripMargin) + val expected = spark.table("parallel_broadcast_build").collect().toSeq + checkAnswer(result, expected) + + val plan = result.queryExecution.executedPlan + assert(plan.collect { case j: BroadcastHashJoinExecTransformer => j }.nonEmpty) + val exchanges = plan.collect { case e: ColumnarBroadcastExchangeExec => e } + assert(exchanges.size == 1) + assert(exchanges.head.metrics("buildThreads").value > 1) Review Comment: The test config forces `broadcastHashTableBuildThreads` to `"1"` but later asserts `buildThreads > 1`, which makes the assertion inconsistent with the configured pool size. To make the test deterministic, either (a) set `broadcastHashTableBuildThreads` to a value > 1 (e.g., 2) and assert `>= 2` (or `== 2` if the metric is expected to match exactly), or (b) stop overriding the thread count in `withSQLConf` and assert based on the default sizing behavior. -- 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]
