[原创] JAVA 递归线程池测试 ExecutorService / ForkJoinPool

测试工具使用递归的方式获取子进程的Msg消息，目前有2种常用的ExecutorService / ForkJoinPool

为了测试哪种效果较好，我们来写个测试Demo，循环5555555次+1（加锁），统计每种执行耗时

int nCpu = Runtime.getRuntime().availableProcessors();

ExecutorService executorPool = Executors.newFixedThreadPool(nCpu);
ForkJoinPool forkJoinPool = new ForkJoinPool(nCpu);

TestData:5555555 , RunTime:1543 ms ：ExecutorService executorPool

TestData:5555555 , RunTime:746 ms ：ForkJoinPool forkJoinPool

结果很明显，递归线程池使用ForkJoinPool更佳，2倍的执行效率

测试流程图

 1 package test;
 2 
 3 import java.util.concurrent.CompletableFuture;
 4 import java.util.concurrent.ExecutorService;
 5 import java.util.concurrent.Executors;
 6 import java.util.concurrent.ForkJoinPool;
 7 
 8 /**
 9  *
10  * @author weimjsam
11  */
12 public class TestThrad {
13 
14     public int addNum = 0;
15 
16     //get cpu 
17     int nCpu = Runtime.getRuntime().availableProcessors();
18 
19     //Thread
20     ExecutorService taskPush = Executors.newFixedThreadPool(nCpu);
21     ExecutorService executorPool = Executors.newFixedThreadPool(nCpu);
22     ForkJoinPool forkJoinPool = new ForkJoinPool(nCpu);
23 
24     private void TaskPush(int iTestAdd) {
25         CompletableFuture.runAsync(() -> {
26 
27             for (int i = 0; i < nCpu; i++) {
28                 CompletableFuture.runAsync(() -> TestRun(iTestAdd), forkJoinPool);
29             }
30 
31         }, taskPush);
32     }
33 
34     private void TestRun(int iTestAdd) {
35         CompletableFuture.runAsync(() -> TestAdd(iTestAdd), forkJoinPool)
36                 .thenRun(() -> CheckOver(iTestAdd));
37     }
38 
39     private void TestAdd(int iTestAdd) {
40         synchronized (this) {
41             if (addNum < iTestAdd) {
42                 addNum = addNum + 1;
43             }
44         }
45     }
46 
47     private void CheckOver(int iTestAdd) {
48         if (addNum < iTestAdd) {
49             TestRun(iTestAdd);
50         }
51     }
52 
53     public void Test(int iTestMax) {
54         TaskPush(iTestMax);
55     }
56 
57 }