mosche commented on code in PR #22182: URL: https://github.com/apache/beam/pull/22182#discussion_r918939204
########## sdks/java/core/jmh/src/main/java/org/apache/beam/sdk/schemas/RowBundle.java: ########## @@ -0,0 +1,240 @@ +/* + * 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.beam.sdk.schemas; + +import static java.util.stream.Collectors.toList; +import static java.util.stream.Collectors.toMap; +import static java.util.stream.IntStream.range; + +import java.nio.charset.StandardCharsets; +import java.util.Map; +import java.util.function.BiConsumer; +import java.util.function.Function; +import java.util.stream.IntStream; +import org.apache.beam.sdk.transforms.SerializableFunction; +import org.apache.beam.sdk.values.Row; +import org.apache.beam.sdk.values.RowWithGetters; +import org.apache.beam.sdk.values.RowWithStorage; +import org.apache.beam.sdk.values.TypeDescriptor; +import org.joda.time.DateTime; +import org.joda.time.Duration; +import org.joda.time.Instant; +import org.openjdk.jmh.annotations.Level; +import org.openjdk.jmh.annotations.Param; +import org.openjdk.jmh.annotations.Scope; +import org.openjdk.jmh.annotations.Setup; +import org.openjdk.jmh.annotations.State; +import org.openjdk.jmh.infra.Blackhole; + +/** + * Bundle of rows according to the configured {@link Factory} as input for benchmarks. + * + * <p>The rows are created during {@link #setup()} to exclude initialization costs from the + * measurement. To prevent unintended cache hits in {@link RowWithGetters}, a new bundle of rows + * must be generated before every invocation. + * + * <p>Setup per {@link Level#Invocation} has considerable drawbacks. Though, given that processing + * bundles of rows (n={@link #bundleSize}) takes well above 1 ms, each individual invocation can be + * adequately timestamped without risking generating wrong results. + */ +@State(Scope.Benchmark) +public class RowBundle { + @SuppressWarnings("ImmutableEnumChecker") // false positive + public enum Action { + /** + * Write field to object using {@link + * GetterBasedSchemaProvider#fromRowFunction(TypeDescriptor)}. + * + * <p>Use {@link RowWithStorage} to bypass optimizations in RowWithGetters for writes. + */ + WRITE(Factory::createWithStorage), + + /** + * Read field from {@link RowWithGetters} provided by {@link + * GetterBasedSchemaProvider#toRowFunction(TypeDescriptor)}. + */ + READ_ONCE(Factory::createWithGetter), + + /** + * Repeatedly (3x) read field from {@link RowWithGetters} provided by {@link Review Comment: I'm not sure I fully understand what you are proposing @lukecwik ... It might be that the terms `read` / `write` are not entirely correct & symmetric in this case: - `read`: invoke getter on the underlying object via `RowWithGetter`(for a single field in this benchmark) - `write`: create new instance of the type represented by `RowWithStorage` and invoke the setter to write the field (again for a single field in this benchmark) (`RowWithGetter` would simply return the underlying object, that's the trivial, cheap path) I think it's valuable to (also) have separate benchmarks for both these to help isolate problems in case of any performance degradation. But having a more use case driven benchmark would certainly be great. Could you elaborate a bit more on the above? -- 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]
