测试Java同步、锁、原子类的同步性能

最新推荐文章于 2022-02-25 00:29:14 发布

本自具足反求诸己

最新推荐文章于 2022-02-25 00:29:14 发布

阅读量198

点赞数

分类专栏： java

本文链接：https://blog.csdn.net/seacean2000/article/details/112285749

版权

java 专栏收录该内容

86 篇文章 0 订阅

订阅专栏

代码：

import java.util.concurrent.TimeUnit;
import java.util.concurrent.atomic.AtomicInteger;
import java.util.concurrent.locks.Lock;
import java.util.concurrent.locks.ReentrantLock;

import org.openjdk.jmh.annotations.Benchmark;
import org.openjdk.jmh.annotations.BenchmarkMode;
import org.openjdk.jmh.annotations.Group;
import org.openjdk.jmh.annotations.GroupThreads;
import org.openjdk.jmh.annotations.Measurement;
import org.openjdk.jmh.annotations.Mode;
import org.openjdk.jmh.annotations.OutputTimeUnit;
import org.openjdk.jmh.annotations.Scope;
import org.openjdk.jmh.annotations.State;
import org.openjdk.jmh.annotations.Warmup;
import org.openjdk.jmh.profile.StackProfiler;
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;
import org.openjdk.jmh.runner.options.TimeValue;

@Measurement(iterations=5)
@Warmup(iterations=5)
@BenchmarkMode(Mode.AverageTime)
@OutputTimeUnit(TimeUnit.MICROSECONDS)
@State(Scope.Group)
public class AtomicTest {
	
	private int x=0;
	private int y=0;
	private AtomicInteger z=new AtomicInteger();
	private final Lock lock=new ReentrantLock();
	
	@GroupThreads(10)
	@Group("lock")
	@Benchmark
	public void lockInc() {
		lock.lock();
		try {
			x++;
		}finally {
			lock.unlock();
		}
	}
	
	@GroupThreads(10)
	@Group("sync")
	@Benchmark
	public void synInc() {
		synchronized(this) {
			y++;
		}
	}
	
	@GroupThreads(10)
	@Group("atom")
	@Benchmark
	public void atomicInc() {
		z.incrementAndGet();
	}
	

	public static void main(String[] args) {
		Options opt=new OptionsBuilder()
				        .include(AtomicTest.class.getSimpleName())
				        .forks(1)
				        .timeout(TimeValue.seconds(10))
				        .addProfiler(StackProfiler.class)
				        .build();
		try {
			new Runner(opt).run();
		} catch (RunnerException e) {
			// TODO Auto-generated catch block
			e.printStackTrace();
		}

	}

}

运行后的测试结果：

Result "jmhtest.AtomicTest.atom":
  0.095 ±(99.9%) 0.005 us/op [Average]
  (min, avg, max) = (0.094, 0.095, 0.098), stdev = 0.001
  CI (99.9%): [0.090, 0.101] (assumes normal distribution)

Secondary result "jmhtest.AtomicTest.atom:·stack":
Stack profiler:

....[Thread state distributions]....................................................................
 89.9%         RUNNABLE
  9.1%         TIMED_WAITING
  1.0%         WAITING

Result "jmhtest.AtomicTest.lock":
  0.216 ±(99.9%) 0.012 us/op [Average]
  (min, avg, max) = (0.214, 0.216, 0.221), stdev = 0.003
  CI (99.9%): [0.205, 0.228] (assumes normal distribution)

Secondary result "jmhtest.AtomicTest.lock:·stack":
Stack profiler:

....[Thread state distributions]....................................................................
 79.2%         WAITING
 11.7%         RUNNABLE
  9.1%         TIMED_WAITING

Result "jmhtest.AtomicTest.sync":
  0.273 ±(99.9%) 0.046 us/op [Average]
  (min, avg, max) = (0.260, 0.273, 0.292), stdev = 0.012
  CI (99.9%): [0.227, 0.318] (assumes normal distribution)

Secondary result "jmhtest.AtomicTest.sync:·stack":
Stack profiler:

....[Thread state distributions]....................................................................
 70.2%         BLOCKED
 20.6%         RUNNABLE
  9.1%         TIMED_WAITING


Benchmark               Mode  Cnt  Score   Error  Units
AtomicTest.atom         avgt    5  0.095 ± 0.005  us/op
AtomicTest.atom:·stack  avgt         NaN            ---
AtomicTest.lock         avgt    5  0.216 ± 0.012  us/op
AtomicTest.lock:·stack  avgt         NaN            ---
AtomicTest.sync         avgt    5  0.273 ± 0.046  us/op
AtomicTest.sync:·stack  avgt         NaN            ---

测试结果分析：

1、从运行时间上来看，原子类的原子性操作运行速度更快，速度是lock同步的2倍以上，速度最慢的synchronized同步。

2、同步运行的时间损耗主要体现在等待和阻塞上，降低阻塞和等待的时间会极大的提高同步的效率。