andygrove commented on code in PR #2979: URL: https://github.com/apache/datafusion-comet/pull/2979#discussion_r2658034465
########## spark/src/test/scala/org/apache/spark/sql/benchmark/CometCastStringToNumericBenchmark.scala: ########## @@ -0,0 +1,95 @@ +/* + * 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.benchmark + +import org.apache.spark.sql.catalyst.expressions.Cast +import org.apache.spark.sql.internal.SQLConf + +import org.apache.comet.CometConf + +case class CastStringToNumericConfig( + name: String, + query: String, + extraCometConfigs: Map[String, String] = Map.empty) + +/** + * Benchmark to measure performance of Comet cast from String to numeric types. To run this + * benchmark: + * {{{ + * SPARK_GENERATE_BENCHMARK_FILES=1 make benchmark-org.apache.spark.sql.benchmark.CometCastStringToNumericBenchmark + * }}} + */ +object CometCastStringToNumericBenchmark extends CometBenchmarkBase { + + private val castFunctions = Seq("CAST", "TRY_CAST") + private val targetTypes = + Seq( + "BOOLEAN", + "BYTE", + "SHORT", + "INT", + "LONG", + "FLOAT", + "DOUBLE", + "DECIMAL(10,2)", + "DECIMAL(38,19)") + + private val castConfigs = for { + castFunc <- castFunctions + targetType <- targetTypes + } yield CastStringToNumericConfig( + s"$castFunc String to $targetType", + s"SELECT $castFunc(c1 AS $targetType) FROM parquetV1Table", + Map( + SQLConf.ANSI_ENABLED.key -> "false", + CometConf.getExprAllowIncompatConfigKey(classOf[Cast]) -> "true")) + + override def runCometBenchmark(mainArgs: Array[String]): Unit = { + val values = 1024 * 1024 // 1M rows + + // Generate input data once for all benchmarks + runBenchmarkWithTable("String to numeric casts", values) { v => + withTempPath { dir => + withTempTable("parquetV1Table") { + // Generate numeric strings with both integer and decimal values + // Also include some special values: nulls (~2%), NaN (~2%), Infinity (~2%) + prepareTable( + dir, + spark.sql(s""" + SELECT CASE + WHEN value % 50 = 0 THEN NULL + WHEN value % 50 = 1 THEN 'NaN' + WHEN value % 50 = 2 THEN 'Infinity' + WHEN value % 50 = 3 THEN '-Infinity' + WHEN value % 50 < 10 THEN CAST(value % 99 AS STRING) + WHEN value % 50 < 30 THEN CAST(value % 999999 AS STRING) + ELSE CAST((value - 500000) / 100.0 AS STRING) + END AS c1 + FROM $tbl + """)) + + castConfigs.foreach { config => + runExpressionBenchmark(config.name, v, config.query, config.extraCometConfigs) Review Comment: yeah, that would be a good feature to add to these microbenchmarks. I filed an issue: https://github.com/apache/datafusion-comet/issues/3019 -- 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]
