【参考来源】：
【1】https://www.jianshu.com/p/c41e942bcd64
【2】https://www.jianshu.com/p/f030aa5d7a28
【3】https://www.cnblogs.com/zhaoyan001/p/7049627.html
【4】https://blog.csdn.net/weixin_37910453/article/details/86523454
【5】https://www.iteye.com/blog/zapldy-746458
文章目录
基础知识
ThreadPoolExecutor的四个构造方法
ThreadPoolExecutor(int corePoolSize, int maximumPoolSize, long keepAliveTime,
TimeUnit unit, BlockingQueue<Runnable> workQueue)
ThreadPoolExecutor(int corePoolSize, int maximumPoolSize, long keepAliveTime,
TimeUnit unit, BlockingQueue<Runnable> workQueue, ThreadFactory threadFactory)
ThreadPoolExecutor(int corePoolSize, int maximumPoolSize, long keepAliveTime, TimeUnit unit,
BlockingQueue<Runnable> workQueue, RejectedExecutionHandler handler)
ThreadPoolExecutor(int corePoolSize, int maximumPoolSize, long keepAliveTime, TimeUnit unit,
BlockingQueue<Runnable> workQueue, RejectedExecutionHandler handler, ThreadFactory threadFactory)
接口参数的含义
| 名称 | 意思 | 类型 |
|---|---|---|
| corePoolSize | 核心线程池的大小 | int |
| maximumPoolSize | 最大线程池大小 | int |
| keepAliveTime | 线程最大空闲时间 | long |
| unit | 时间单位 | TimeUnit |
| workQueue | 线程等待队列 | BlockingQueue < Runnable > |
| threadFactory | 线程创建工厂 | ThreadFactory |
| handler | 拒绝策略 | RejectedExecutionHandler |
线程池实战
【例1】循环100次统计一个数值，每次计算累加的操作耗时 100毫秒。单线程和多线程的做法。
public static void no() throws Exception {
long start = System.currentTimeMillis();
Integer count = 100;
Integer sum = 0;
for (int a = 0; a < count; a++) {
sum += a;
Thread.sleep(100);
}
System.out.println(String.format("结果：%d\n耗时：%d", sum, System.currentTimeMillis() - start));
}
public static void yes() throws Exception {
long start = System.currentTimeMillis();
Integer count = 100;
// CountDownLatch 在完成最后一次计算后通知程序已经计算完成，节省等待时间
// CountDownLatch其实可以把它看作一个计数器，只不过这个计数器的操作是原子操作，
// 同时只能有一个线程去操作这个计数器，
// 也就是同时只能有一个线程去减这个计数器里面的值。
CountDownLatch countDownLatch = new CountDownLatch(count);
// 使用AtomicInteger，线程是安全的，保证了数据正确性
AtomicInteger sum = new AtomicInteger(0);
for (int a = 0; a < count; a++) {
final Integer e = a;
new Thread(() -> {
sum.getAndAdd(e);
try {
Thread.sleep(100);
} catch (InterruptedException e1) {
e1.printStackTrace();
}
countDownLatch.countDown();
}).start();
}
countDownLatch.await();
System.out.println(String.format("结果：%d\n耗时：%d", sum.get(), System.currentTimeMillis() - start));
}
public static void main(String[] args) throws Exception {
System.out.println("普通做法");
no();
System.out.println("多线程做法");
yes();
}
【运行结果】
【例2】三个工人为老板干活，当三个工人把一天的活都干完了的时候，老板就来检查所有工人所干的活。
两个实体类，Worker代表工人，Boss代表老板。
CountDownLatch的一个非常典型的应用场景是：有一个任务想要往下执行，但必须要等到其他的任务执行完毕后才可以继续往下执行。假如我们这个想要继续往下执行的任务调用一个CountDownLatch对象的await()方法，其他的任务执行完自己的任务后调用同一个CountDownLatch对象上的countDown()方法，这个调用await()方法的任务将一直阻塞等待，直到这个CountDownLatch对象的计数值减到0为止。
Atomiclong( ) 可以理解是加了synchronized的long。
【Woker类】
// Worker类
import java.util.Random;
import java.util.concurrent.CountDownLatch;
import java.util.concurrent.TimeUnit;
public class Worker implements Runnable {
private CountDownLatch cdl;
private String name;
public Worker(CountDownLatch cdl, String name) {
this.cdl = cdl;
this.name = name;
}
public void run() {
this.doWork();
try {
TimeUnit.SECONDS.sleep(new Random().nextInt(10));
} catch (InterruptedException ie) {
}
System.out.println(this.name + "活干完了！");
this.cdl.countDown();
}
private void doWork() {
System.out.println(this.name + "正在干活!");
}
}
【Boss类】
// Boss类
import java.util.concurrent.CountDownLatch;
public class Boss implements Runnable {
private CountDownLatch downLatch;
public Boss(CountDownLatch downLatch){
this.downLatch = downLatch;
}
public void run() {
System.out.println("老板正在等所有的工人干完活......");
try {
this.downLatch.await();
} catch (InterruptedException e) {
}
System.out.println("工人活都干完了，老板开始检查了！");
}
}
【主类】
import java.util.concurrent.CountDownLatch;
import java.util.concurrent.ExecutorService;
import java.util.concurrent.Executors;
public class Demo {
public static void main(String[] args) {
ExecutorService executor = Executors.newCachedThreadPool();
CountDownLatch cdl = new CountDownLatch(3);
Worker w1 = new Worker(cdl, "张三");
Worker w2 = new Worker(cdl, "李四");
Worker w3 = new Worker(cdl, "王二");
Boss boss = new Boss(cdl);
executor.execute(w3);
executor.execute(w2);
executor.execute(w1);
executor.execute(boss);
executor.shutdown();
}
}
【运行结果】
【知识】：运行主类的时候，可以肯定的每次运行的结果可能与图的不一样，但老板检查永远是在后面的。