简单的 JMH 示例

编写适当的基准测试的工具之一是 JMH 。假设我们要比较在 HashSetTreeSet 中搜索元素的性能。

将 JHM 引入项目的最简单方法是使用 maven 和 shade 插件。你也可以从 JHM 示例中看到 pom.xml

<build>
    <plugins>
        <plugin>
            <groupId>org.apache.maven.plugins</groupId>
            <artifactId>maven-shade-plugin</artifactId>
            <version>3.0.0</version>
            <executions>
                <execution>
                    <phase>package</phase>
                    <goals>
                        <goal>shade</goal>
                    </goals>
                    <configuration>
                        <finalName>/benchmarks</finalName>
                        <transformers>
                            <transformer
                                    implementation="org.apache.maven.plugins.shade.resource.ManifestResourceTransformer">
                                <mainClass>org.openjdk.jmh.Main</mainClass>
                            </transformer>
                        </transformers>
                        <filters>
                            <filter>
                                <artifact>*:*</artifact>
                                <excludes>
                                    <exclude>META-INF/*.SF</exclude>
                                    <exclude>META-INF/*.DSA</exclude>
                                    <exclude>META-INF/*.RSA</exclude>
                                </excludes>
                            </filter>
                        </filters>
                    </configuration>
                </execution>
            </executions>
        </plugin>
    </plugins>
</build>

<dependencies>
    <dependency>
        <groupId>org.openjdk.jmh</groupId>
        <artifactId>jmh-core</artifactId>
        <version>1.18</version>
    </dependency>
    <dependency>
        <groupId>org.openjdk.jmh</groupId>
        <artifactId>jmh-generator-annprocess</artifactId>
        <version>1.18</version>
    </dependency>
</dependencies>

在此之后,你需要编写基准类本身:

package benchmark;

import org.openjdk.jmh.annotations.*;
import org.openjdk.jmh.infra.Blackhole;

import java.util.HashSet;
import java.util.Random;
import java.util.Set;
import java.util.TreeSet;
import java.util.concurrent.TimeUnit;

@State(Scope.Thread)
public class CollectionFinderBenchmarkTest {
    private static final int SET_SIZE = 10000;

    private Set<String> hashSet;
    private Set<String> treeSet;

    private String stringToFind = "8888";

    @Setup
    public void setupCollections() {
        hashSet = new HashSet<>(SET_SIZE);
        treeSet = new TreeSet<>();

        for (int i = 0; i < SET_SIZE; i++) {
            final String value = String.valueOf(i);
            hashSet.add(value);
            treeSet.add(value);
        }

        stringToFind = String.valueOf(new Random().nextInt(SET_SIZE));
    }

    @Benchmark
    @BenchmarkMode(Mode.AverageTime)
    @OutputTimeUnit(TimeUnit.NANOSECONDS)
    public void testHashSet(Blackhole blackhole) {
        blackhole.consume(hashSet.contains(stringToFind));
    }

    @Benchmark
    @BenchmarkMode(Mode.AverageTime)
    @OutputTimeUnit(TimeUnit.NANOSECONDS)
    public void testTreeSet(Blackhole blackhole) {
        blackhole.consume(treeSet.contains(stringToFind));
    }
}

请记住这个 blackhole.consume(),我们稍后再回过头来看看。我们还需要运行基准的主类:

package benchmark;

import org.openjdk.jmh.runner.Runner;
import org.openjdk.jmh.runner.RunnerException;
import org.openjdk.jmh.runner.options.Options;
import org.openjdk.jmh.runner.options.OptionsBuilder;

public class BenchmarkMain {
    public static void main(String[] args) throws RunnerException {
        final Options options = new OptionsBuilder()
                .include(CollectionFinderBenchmarkTest.class.getSimpleName())
                .forks(1)
                .build();

        new Runner(options).run();
    }
}

我们都准备好了。我们只需要运行 mvn package(它将在你的/target 文件夹中创建 benchmarks.jar)并运行我们的基准测试:

java -cp target/benchmarks.jar benchmark.BenchmarkMain

经过一些预热和计算迭代后,我们将获得结果:

# Run complete. Total time: 00:01:21

Benchmark                                  Mode  Cnt   Score    Error  Units
CollectionFinderBenchmarkTest.testHashSet  avgt   20   9.940 ±  0.270  ns/op
CollectionFinderBenchmarkTest.testTreeSet  avgt   20  98.858 ± 13.743  ns/op

关于那个 blackhole.consume()。如果你的计算不会改变应用程序的状态,那么 java 很可能会忽略它。因此,为了避免它,你可以使你的基准测试方法返回一些值,或使用 Blackhole 对象来使用它。

你可以找到关于写适当的基准测试的更多信息阿列克谢 Shipilëv 的博客 ,在雅各 Jenkov 的博客和 Java 的性能博客: 12