HotSpot中Serial/Serial Old与Parallel Scavenge/Parallel Old内存分配策略区别

按照《深入理解Java虚拟机》中测试内存分配与回收策略中的代码进行了测试，

JDK8官方文档中JVM相关的文档地址https://docs.oracle.com/javase/8/docs/technotes/guides/vm/gctuning/

测试代码如下，使用的是JDK8，默认的垃圾收集器是Parallel Scavenge+Parallel Old

public class TestAllocation {

	private static final int _1MB = 1024 * 1024;

	public static void main(String[] args) {
		byte[] allocation1, allocation2, allocation3, allocation4;
		allocation1 = new byte[2 * _1MB];
		allocation1 = new byte[2 * _1MB];
		allocation1 = new byte[2 * _1MB];
		allocation1 = new byte[4 * _1MB];
	}
}

在使用Serial/Serial Old 垃圾收集器的情况下测试结果如下：

[root@iZbp13pwlxqwiu1xxb6szsZ com]# java -verbose:gc -Xms20M -Xmx20M -Xmn10M -XX:+PrintGCDetails -XX:SurvivorRatio=8 -XX:+UseSerialGC TestAllocation
[GC (Allocation Failure) [DefNew: 6815K->259K(9216K), 0.0030945 secs] 6815K->2307K(19456K), 0.0031645 secs] [Times: user=0.00 sys=0.00, real=0.00 secs] 
Heap
 def new generation   total 9216K, used 4437K [0x00000000fec00000, 0x00000000ff600000, 0x00000000ff600000)
  eden space 8192K,  51% used [0x00000000fec00000, 0x00000000ff014930, 0x00000000ff400000)
  from space 1024K,  25% used [0x00000000ff500000, 0x00000000ff540c98, 0x00000000ff600000)
  to   space 1024K,   0% used [0x00000000ff400000, 0x00000000ff400000, 0x00000000ff500000)
 tenured generation   total 10240K, used 2048K [0x00000000ff600000, 0x0000000100000000, 0x0000000100000000)
   the space 10240K,  20% used [0x00000000ff600000, 0x00000000ff800010, 0x00000000ff800200, 0x0000000100000000)
 Metaspace       used 2492K, capacity 4486K, committed 4864K, reserved 1056768K
  class space    used 267K, capacity 386K, committed 512K, reserved 1048576K
[root@iZbp13pwlxqwiu1xxb6szsZ com]# 

 发生一次Minor GC，新生代空间由6815K变成了259K，

在使用Parallel Scavenge/Parallel Old垃圾收集器测试结果如下：

[root@iZbp13pwlxqwiu1xxb6szsZ com]# java -verbose:gc -Xms20M -Xmx20M -Xmn10M -XX:+PrintGCDetails -XX:SurvivorRatio=8  TestAllocation
Heap
 PSYoungGen      total 9216K, used 6980K [0x00000000ff600000, 0x0000000100000000, 0x0000000100000000)
  eden space 8192K, 85% used [0x00000000ff600000,0x00000000ffcd1020,0x00000000ffe00000)
  from space 1024K, 0% used [0x00000000fff00000,0x00000000fff00000,0x0000000100000000)
  to   space 1024K, 0% used [0x00000000ffe00000,0x00000000ffe00000,0x00000000fff00000)
 ParOldGen       total 10240K, used 4096K [0x00000000fec00000, 0x00000000ff600000, 0x00000000ff600000)
  object space 10240K, 40% used [0x00000000fec00000,0x00000000ff000010,0x00000000ff600000)
 Metaspace       used 2491K, capacity 4486K, committed 4864K, reserved 1056768K
  class space    used 267K, capacity 386K, committed 512K, reserved 1048576K
[root@iZbp13pwlxqwiu1xxb6szsZ com]# 

新生代剩余的空间已经不足以存放allocation4对象，该对象直接被分配在老年代，此时并不会发生GC。

大对象直接进入老年代

我们直接分配一个4M的对象

public class TestAllocation {
	private static final int _1MB = 1024 * 1024;
	public static void main(String[] args) {
		byte[] allocation1;
		allocation1 = new byte[4 * _1MB];
	}
}

虚拟机提供了一个-XX:PretenureSizeThreshold参数，令大于这个设置值的对象直接在老年代分配。这样做的目的是避免在Eden区及两个Survivor区之间发生大量的内存复制。

我们进行测试如下： 

[root@iZbp13pwlxqwiu1xxb6szsZ com]# java -Xms20M -Xmx20M -Xmn10M -XX:+PrintGCDetails -XX:SurvivorRatio=8 -XX:+UseSerialGC TestAllocation
Heap
 def new generation   total 9216K, used 4932K [0x00000000fec00000, 0x00000000ff600000, 0x00000000ff600000)
  eden space 8192K,  60% used [0x00000000fec00000, 0x00000000ff0d1000, 0x00000000ff400000)
  from space 1024K,   0% used [0x00000000ff400000, 0x00000000ff400000, 0x00000000ff500000)
  to   space 1024K,   0% used [0x00000000ff500000, 0x00000000ff500000, 0x00000000ff600000)
 tenured generation   total 10240K, used 0K [0x00000000ff600000, 0x0000000100000000, 0x0000000100000000)
   the space 10240K,   0% used [0x00000000ff600000, 0x00000000ff600000, 0x00000000ff600200, 0x0000000100000000)
 Metaspace       used 2491K, capacity 4486K, committed 4864K, reserved 1056768K
  class space    used 267K, capacity 386K, committed 512K, reserved 1048576K
[root@iZbp13pwlxqwiu1xxb6szsZ com]# java -Xms20M -Xmx20M -Xmn10M -XX:+PrintGCDetails -XX:SurvivorRatio=8 -XX:+UseSerialGC -XX:PretenureSizeThreshold=3145728 TestAllocation
Heap
 def new generation   total 9216K, used 835K [0x00000000fec00000, 0x00000000ff600000, 0x00000000ff600000)
  eden space 8192K,  10% used [0x00000000fec00000, 0x00000000fecd0ff0, 0x00000000ff400000)
  from space 1024K,   0% used [0x00000000ff400000, 0x00000000ff400000, 0x00000000ff500000)
  to   space 1024K,   0% used [0x00000000ff500000, 0x00000000ff500000, 0x00000000ff600000)
 tenured generation   total 10240K, used 4096K [0x00000000ff600000, 0x0000000100000000, 0x0000000100000000)
   the space 10240K,  40% used [0x00000000ff600000, 0x00000000ffa00010, 0x00000000ffa00200, 0x0000000100000000)
 Metaspace       used 2491K, capacity 4486K, committed 4864K, reserved 1056768K
  class space    used 267K, capacity 386K, committed 512K, reserved 1048576K
[root@iZbp13pwlxqwiu1xxb6szsZ com]# 

当没有使用-XX:PretenureSizeThreshold参数时，一个4M的对象分配到了Eden区，当使用了-XX:PretenureSizeThreshold=3145728，(就是3M，该参数不能直接写成3M)时，这个对象直接被分配到了老年代。

扫描二维码关注公众号，回复： 9199837 查看本文章

注意：PretenureSizeThreshold参数只对Serial和ParNew两款收集器有效，ParallelScavenge收集器不认识这个参数，ParallelScavenge收集器一般并不需要设置。如果遇到必须使用此参数的场合，可以考虑ParNew加CMS的收集器组合。