Java8中用sun.misc.Contended避免伪共享(false sharing)

  关于伪共享这个概念，请先参照http://ifeve.com/falsesharing/

伪共享的样子：

Java view:
public class VolatileLong {
		volatile long v = 0L;
}

Memory view:
...–––-)(––––HV––––HV–––)(–––...

我们看到，两个VolatileLong对象被load到了同一个缓存行里面，如果一个线程要修改对象1，另一个线程同时要修改对象2，此时就要面对伪共享这个无形的性能杀手了

jdk6中的解决办法：

Java view:
public class VolatileLong {
		volatile long p0, p1, p2, p3, p4, p5, p6;
		volatile long v = 0L;
		volatile long q0, q1, q2, q3, q4, q5, q6;
}

Memory view:
...–––-)(ppppppHVqqqqqq)(–––...

很多大神的代码都是通过上面的方式也就是long padding来避免伪共享
例如：
1.Doug Lea的jsr166中早期的LinkedTransferQueue版本http://gee.cs.oswego.edu/cgi-bin/viewcvs.cgi/jsr166/src/jsr166y/LinkedTransferQueue.java?revision=1.1&view=markup
2.还是Doug Lea的ConcurrentHashMapV8http://gee.cs.oswego.edu/cgi-bin/viewcvs.cgi/jsr166/src/jsr166e/ConcurrentHashMapV8.java?revision=1.121&view=markup（在java8中的版本已经用sun.misc.Contended替换long padding）
3.大名鼎鼎的无锁并发框架Disruptor https://github.com/LMAX-Exchange/disruptor
4.等等等...

long padding的解决办法不怎么优雅，并且在jdk7某个版本以后能会优化掉long padding，尼玛java程序员是有多难啊，想尽了一切办法对付jdk7中的这个优化，详情点击http://ifeve.com/false-sharing-java-7/

但是但是...开始使用jdk8的同学注意了，java8已经给出了官方的解决办法...
就是sun.misc.Contended注解http://mail.openjdk.java.net/pipermail/hotspot-dev/2012-November/007309.html

Java view:
// jdk8新特性，Contended注解避免false sharing
// Restricted on user classpath
// Unlock: -XX:-RestrictContended
@sun.misc.Contended
public class VolatileLong {
		volatile long v = 0L;
}


Memory view:
...–––-)(******HV******)(–––...

要注意的是user classpath使用此注解默认是无效的，需要在jvm启动时设置-XX:-RestrictContended

jdk8中已经使用sun.misc.Contended的地方：

src/share/classes/java/util/concurrent/ConcurrentHashMap.java
2458: @sun.misc.Contended static final class CounterCell {

src/share/classes/java/util/concurrent/Exchanger.java
310: * bookkeeping. Padded via @sun.misc.Contended to reduce m
313: @sun.misc.Contended static final class Node {

src/share/classes/java/util/concurrent/ForkJoinPool.java 161:@sun.misc.Contended
643: @sun.misc.Contended

src/share/classes/java/util/concurrent/atomic/Striped64.java
55: * (via @sun.misc.Contended) to reduce cache contention. Pa
119: @sun.misc.Contended static final class Cell {

src/share/classes/java/lang/Thread.java
2004: @sun.misc.Contended("tlr")
2008: @sun.misc.Contended("tlr")
2012: @sun.misc.Contended("tlr")

最后，贴上测试代码，感兴趣的各自测试吧

public class FalseSharing implements Runnable {

	public final static int NUM_THREADS = 4; // change
	public final static long ITERATIONS = 500L * 1000L * 1000L;
	private final int arrayIndex;

	private static VolatileLong3[] longs = new VolatileLong3[NUM_THREADS];
	static {
		for (int i = 0; i < longs.length; i++) {
			longs[i] = new VolatileLong3();
		}
	}

	public FalseSharing(final int arrayIndex) {
		this.arrayIndex = arrayIndex;
	}

	public static void main(final String[] args) throws Exception {
		long start = System.nanoTime();
		runTest();
		System.out.println("duration = " + (System.nanoTime() - start));
	}

	private static void runTest() throws InterruptedException {
		Thread[] threads = new Thread[NUM_THREADS];

		for (int i = 0; i < threads.length; i++) {
			threads[i] = new Thread(new FalseSharing(i));
		}

		for (Thread t : threads) {
			t.start();
		}

		for (Thread t : threads) {
			t.join();
		}
	}

	public void run() {
		long i = ITERATIONS + 1;
		while (0 != --i) {
			longs[arrayIndex].value = i;
		}
	}

	public final static class VolatileLong {
		public volatile long value = 0L;
	}

	// long padding避免false sharing
	// 按理说jdk7以后long padding应该被优化掉了，但是从测试结果看padding仍然起作用
	public final static class VolatileLong2 {
		volatile long p0, p1, p2, p3, p4, p5, p6;
		public volatile long value = 0L;
		volatile long q0, q1, q2, q3, q4, q5, q6;
	}

	// jdk8新特性，Contended注解避免false sharing
	// Restricted on user classpath
	// Unlock: -XX:-RestrictContended
	@sun.misc.Contended
	public final static class VolatileLong3 {
		public volatile long value = 0L;
	}
}

我机器上的测试结果：

VolatileLong： duration = 29594765000
VolatileLong2：duration = 7234555000
VolatileLong3：duration = 7167475000

测试代码来自http://ifeve.com/falsesharing/，稍有改动

评论

2 楼 budairenqin 2014-09-15  

liuInsect 写道

      volatile long p0, p1, p2, p3, p4, p5, p6; 
        public volatile long value = 0L; 
        volatile long q0, q1, q2, q3, q4, q5, q6; 
这里 为什么是 13 * 8 个字节 还不是 （7+1）*8？  8*8 =64不是正好是一个cache line么？

是15*8，因为用7个下标保证了会有56个字节填充在数值的任何一边，56字节的填充+8字节的long数值正好装进一行64字节的缓存行。

1 楼 liuInsect 2014-05-21  

      volatile long p0, p1, p2, p3, p4, p5, p6; 
        public volatile long value = 0L; 
        volatile long q0, q1, q2, q3, q4, q5, q6; 
这里 为什么是 13 * 8 个字节 还不是 （7+1）*8？  8*8 =64不是正好是一个cache line么？