细数Android开源项目中那些频繁使用的并发库中的类

这篇blog旨在帮助大家 梳理一下前面分析的那些开源代码中喜欢使用的一些类，这对我们真正理解这些项目是有极大好处的，以后遇到类似问题 我们就可以自己模仿他们也写

出类似的代码。

1.ExecutorService

这个类实际上就是一个接口

1 public interface ExecutorService extends Executor {

我们可以看看有哪些频繁使用的类 是实现了这个接口的，其实主要就是3个。

 1  /**  2      * Creates a thread pool that reuses a fixed number of threads  3      * operating off a shared unbounded queue.  At any point, at most  4      * {@code nThreads} threads will be active processing tasks.  5      * If additional tasks are submitted when all threads are active,  6      * they will wait in the queue until a thread is available.  7      * If any thread terminates due to a failure during execution  8      * prior to shutdown, a new one will take its place if needed to  9      * execute subsequent tasks.  The threads in the pool will exist 10      * until it is explicitly {@link ExecutorService#shutdown shutdown}. 11      * 12      * @param nThreads the number of threads in the pool 13      * @return the newly created thread pool 14      * @throws IllegalArgumentException if {@code nThreads <= 0} 15      */ 16     public static ExecutorService newFixedThreadPool(int nThreads) { 17         return new ThreadPoolExecutor(nThreads, nThreads, 18                                       0L, TimeUnit.MILLISECONDS, 19                                       new LinkedBlockingQueue<Runnable>()); 20     }

这个线程池，就是有固定线程数的一个线程池，有共享的无界队列来运行这些线程。

 1 /**  2      * Creates a thread pool that creates new threads as needed, but  3      * will reuse previously constructed threads when they are  4      * available.  These pools will typically improve the performance  5      * of programs that execute many short-lived asynchronous tasks.  6      * Calls to {@code execute} will reuse previously constructed  7      * threads if available. If no existing thread is available, a new  8      * thread will be created and added to the pool. Threads that have  9      * not been used for sixty seconds are terminated and removed from 10      * the cache. Thus, a pool that remains idle for long enough will 11      * not consume any resources. Note that pools with similar 12      * properties but different details (for example, timeout parameters) 13      * may be created using {@link ThreadPoolExecutor} constructors. 14      * 15      * @return the newly created thread pool 16      */ 17     public static ExecutorService newCachedThreadPool() { 18         return new ThreadPoolExecutor(0, Integer.MAX_VALUE, 19                                       60L, TimeUnit.SECONDS, 20                                       new SynchronousQueue<Runnable>()); 21     }

这个线程池，是根据需要来创建这些线程的，但是以前构造过的线程 必要时可以重用他们，所以这个在很多android的开源项目里都有用到，很频繁，对于执行很多短期的异步任务来说，这个线程池可以极大的提高程序的性能。

 1 /**  2      * Creates an Executor that uses a single worker thread operating  3      * off an unbounded queue. (Note however that if this single  4      * thread terminates due to a failure during execution prior to  5      * shutdown, a new one will take its place if needed to execute  6      * subsequent tasks.)  Tasks are guaranteed to execute  7      * sequentially, and no more than one task will be active at any  8      * given time. Unlike the otherwise equivalent  9      * {@code newFixedThreadPool(1)} the returned executor is 10      * guaranteed not to be reconfigurable to use additional threads. 11      * 12      * @return the newly created single-threaded Executor 13      */ 14     public static ExecutorService newSingleThreadExecutor() { 15         return new FinalizableDelegatedExecutorService 16             (new ThreadPoolExecutor(1, 1, 17                                     0L, TimeUnit.MILLISECONDS, 18                                     new LinkedBlockingQueue<Runnable>())); 19     }

而这个线程池就比较特殊一点，他只有一个worker线程在工作。

来看第一个程序：

 1 public class Test1 {  2   3     public static void main(String[] args) {  4         ExecutorService exectrorService = Executors.newFixedThreadPool(10);  5         // execute异步的方法去执行这个runnable 但是这种方法无法取得运行之后的返回值  6         exectrorService.execute(new Runnable() {  7             @Override  8             public void run() {  9                 // TODO Auto-generated method stub 10                 int i = 0; 11                 while (true) { 12                     try { 13                         Thread.sleep(2000); 14                     } catch (InterruptedException e) { 15                         // TODO Auto-generated catch block 16                         e.printStackTrace(); 17                     } 18                     System.out.println(i); 19                     i++; 20                 } 21             } 22  23         }); 24  25         exectrorService.execute(new Runnable() { 26             @Override 27             public void run() { 28                 // TODO Auto-generated method stub 29                 int i = 100; 30                 while (true) { 31                     try { 32                         Thread.sleep(2000); 33                     } catch (InterruptedException e) { 34                         // TODO Auto-generated catch block 35                         e.printStackTrace(); 36                     } 37                     System.out.println(i); 38                     i++; 39                 } 40             } 41  42         });

很简单 没有什么好说的只是为了演示一下这个方法，继续往下看：

 1 public class Test1 {  2   3     public static void main(String[] args) {  4         ExecutorService exectrorService = Executors.newFixedThreadPool(10);  5         Future future = exectrorService.submit(new Runnable() {  6   7             @Override  8             public void run() {  9                 System.out.println("thread start"); 10                 // TODO Auto-generated method stub 11                 try { 12                     Thread.sleep(13000); 13                 } catch (InterruptedException e) { 14                     // TODO Auto-generated catch block 15                     e.printStackTrace(); 16                 } 17                 System.out.println("task done"); 18             } 19         }); 20         System.out.println("ready to print status"); 21         try { 22             // 执行完毕以后才会返回null,如果线程还没有执行完毕 那这个地方会阻塞 23             System.out.println("future.get ==" + future.get()); 24         } catch (InterruptedException e) { 25             // TODO Auto-generated catch block 26             e.printStackTrace(); 27         } catch (ExecutionException e) { 28             // TODO Auto-generated catch block 29             e.printStackTrace(); 30         } 31         System.out.println("finish ready");

这个就是为了演示get方法是个阻塞方法的 我们可以看下打印的日志。

程序一开始运行 日志如下：

thread startready to print status

当线程执行完毕大约过了13秒以后

才会继续输入日志如下：

继续看下面的例子：

 1 package com.android.testclass;  2   3 import java.util.concurrent.Callable;  4 import java.util.concurrent.ExecutionException;  5 import java.util.concurrent.ExecutorService;  6 import java.util.concurrent.Executors;  7 import java.util.concurrent.Future;  8   9 public class Test1 { 10  11     public static void main(String[] args) { 12         ExecutorService exectrorService = Executors.newFixedThreadPool(10); 13         // 这个submit方法则会保证结束以后把结果返回给future，用泛型定义的方法 你可以 14         // 用任意的object代替T 15         Future future = exectrorService.submit(new Callable<String>() { 16             @Override 17             public String call() throws Exception { 18                 // TODO Auto-generated method stub 19                 System.out.println("call start"); 20  21                 Thread.sleep(5000); 22  23                 return "call done"; 24             } 25         }); 26         System.out.println("ready to print"); 27         try { 28             System.out.println("future.get()" + future.get()); 29         } catch (InterruptedException e) { 30             // TODO Auto-generated catch block 31             e.printStackTrace(); 32         } catch (ExecutionException e) { 33             // TODO Auto-generated catch block 34             e.printStackTrace(); 35         } 36         System.out.println("finish"); 37  38     } 39 }

同样是submit方法 只不过这次我们换了一个参数 这次是callable参数，这么做的好处就是执行完毕以后可以拿到结果了

一开始输出：

call startready to print

线程执行完毕以后输出：

future.get()call donefinish

然后我们继续看invokeany这个函数：

 1 package com.android.testclass;  2   3 import java.util.HashSet;  4 import java.util.Set;  5 import java.util.concurrent.Callable;  6 import java.util.concurrent.ExecutionException;  7 import java.util.concurrent.ExecutorService;  8 import java.util.concurrent.Executors;  9  10 public class Test2 { 11  12     public static void main(String[] args) { 13         ExecutorService executorService = Executors.newFixedThreadPool(10); 14         Set<Callable<String>> callables = new HashSet<Callable<String>>(); 15         callables.add(new Callable<String>() { 16             @Override 17             public String call() throws Exception { 18                 // TODO Auto-generated method stub 19                 System.out.println("task 1 start"); 20                 Thread.sleep(3000); 21                 return "Task 1"; 22             } 23         }); 24         callables.add(new Callable<String>() { 25             @Override 26             public String call() throws Exception { 27                 System.out.println("task 2 start"); 28                 Thread.sleep(3000); 29                 return "Task 2"; 30             } 31         }); 32         callables.add(new Callable<String>() { 33             @Override 34             public String call() throws Exception { 35                 System.out.println("task 3 start"); 36                 Thread.sleep(3000); 37                 return "Task 3"; 38             } 39         }); 40         System.out.println("ready to print"); 41         try { 42             //返回某一个callable执行结束的结果，结果并不确定 43             String result = executorService.invokeAny(callables); 44             System.out.println("result==" + result); 45         } catch (InterruptedException e) { 46             // TODO Auto-generated catch block 47             e.printStackTrace(); 48         } catch (ExecutionException e) { 49             // TODO Auto-generated catch block 50             e.printStackTrace(); 51         } 52         System.out.println("done to print"); 53  54     } 55 }

输出我就不放了 大家可以自己跑一下。这个函数用的比较少。

那下面这个invokeall函数用的就比较多了

 1 package com.android.testclass;  2   3 import java.util.HashSet;  4 import java.util.List;  5 import java.util.Set;  6 import java.util.concurrent.Callable;  7 import java.util.concurrent.ExecutionException;  8 import java.util.concurrent.ExecutorService;  9 import java.util.concurrent.Executors; 10 import java.util.concurrent.Future; 11  12 public class Test3 { 13  14     public static void main(String[] args) { 15         ExecutorService executorService = Executors.newFixedThreadPool(10); 16         Set<Callable<String>> callables = new HashSet<Callable<String>>(); 17         callables.add(new Callable<String>() { 18             @Override 19             public String call() throws Exception { 20                 // TODO Auto-generated method stub 21                 System.out.println("task 1 start"); 22                 Thread.sleep(3000); 23                 return "Task 1"; 24             } 25         }); 26         callables.add(new Callable<String>() { 27             @Override 28             public String call() throws Exception { 29                 System.out.println("task 2 start"); 30                 Thread.sleep(6000); 31                 return "Task 2"; 32             } 33         }); 34         callables.add(new Callable<String>() { 35             @Override 36             public String call() throws Exception { 37                 System.out.println("task 3 start"); 38                 Thread.sleep(9000); 39                 return "Task 3"; 40             } 41         }); 42         System.out.println("ready to print"); 43  44         try { 45             // invoke方法也是阻塞方法，一定是所有callable都执行完毕才会返回结果 46             List<Future<String>> futures = executorService.invokeAll(callables); 47             System.out.println("invoke done"); 48             for (Future<String> future : futures) { 49                 System.out.println("future.get=" + future.get()); 50                 System.out.println("get done"); 51             } 52             System.out.println("all get done"); 53         } catch (InterruptedException e) { 54             // TODO Auto-generated catch block 55             e.printStackTrace(); 56         } catch (ExecutionException e) { 57             // TODO Auto-generated catch block 58             e.printStackTrace(); 59         } 60  61     } 62 }

总的来说，在android里如果你要使用线程池的话，那上面的这些方法 基本就肯定足够你使用了。

2.ConcurrentHashMap

这个类，相信很多人都不陌生，我就略微提一下，很多人以前在单线程的时候使用hashmap，多线程的时候使用hashtable，这么做虽然是对的，

但是hashtable里的源码说明了 这是直接对整个map进行加锁，效率是很低的，而这个concurrenthashmap的读操作几乎不会有锁，

而写操作由于采用了分段处理，所以写操作的锁 的概率和次数也大大降低。总体来说这是一个效率极高的 可适用于并发性的hashmap。

例子和原理 网上有很多 我这里就不放了。

此外和他类似的还有LinkedHashMap，实现LRU的最好选择，这个也不多讲，只是提一下，网上资料很多。

3.PriorityBlockingQueue

这个就是优先级队列，当然也是支持并发的，这个队列里存放的对象 必须是实现了Comparable 接口的。并且小的是在这个队列前面的 大的就一定是在队列的后面。

比如说我们先定义一个类：

 1 package com.android.testclass;  2   3 public class PriorityEntity implements Comparable<PriorityEntity> {  4   5     private static int count = 0;  6     private int id = count++;  7     private int priority;  8     private int index = 0;  9  10     public PriorityEntity(int priority, int index) { 11         // TODO Auto-generated constructor stub 12         this.priority = priority; 13         this.index = index; 14     } 15  16     @Override 17     public String toString() { 18         return "PriorityEntity [id=" + id + ", priority=" + priority + ", index=" + index + "]"; 19     } 20  21     @Override 22     public int compareTo(PriorityEntity o) { 23         // TODO Auto-generated method stub 24         return this.priority > o.priority ? 1 : this.priority < o.priority ? -1 : 0; 25     } 26  27 }

那个静态变量就表示索引的，构造出一个对象 索引就加1. 然后我们来写一下测试这个队列的代码：

 1 package com.android.testclass;  2   3 import java.util.Random;  4 import java.util.concurrent.ExecutorService;  5 import java.util.concurrent.Executors;  6 import java.util.concurrent.PriorityBlockingQueue;  7 import java.util.concurrent.TimeUnit;  8   9 public class Test6 { 10  11     public static void main(String[] args) { 12         // TODO Auto-generated method stub 13  14         PriorityBlockingQueue q = new PriorityBlockingQueue<>(); 15         Random r = new Random(47); 16         ExecutorService se = Executors.newCachedThreadPool(); 17         //往队列里 放对象,priority的值是 随即的 18         se.execute(new Runnable() { 19  20             @Override 21             public void run() { 22                 // TODO Auto-generated method stub 23                 int i = 0; 24                 while (true) { 25                     q.put(new PriorityEntity(r.nextInt(10), i++)); 26  27                     try { 28                         TimeUnit.MILLISECONDS.sleep(r.nextInt(1000)); 29                     } catch (InterruptedException e) { 30                         // TODO Auto-generated catch block 31                         e.printStackTrace(); 32                     } 33  34                 } 35             } 36         }); 37         //从队列里 取对象,然后把队列里剩余的值打出来 就会发现 每次取出来的都是最小的那个 剩下的都是从小到大排序好的 38         se.execute(new Runnable() { 39  40             @Override 41             public void run() { 42                 // TODO Auto-generated method stub 43                 while (true) { 44                     try { 45                         System.out.println(("take-- " + q.take() + " left:-- [" + q.toString() + "]")); 46                     } catch (InterruptedException e1) { 47                         // TODO Auto-generated catch block 48                         e1.printStackTrace(); 49                     } 50                     try { 51                         TimeUnit.MILLISECONDS.sleep(r.nextInt(1000)); 52                     } catch (InterruptedException e) { 53                         // TODO Auto-generated catch block 54                         e.printStackTrace(); 55                     } 56  57                 } 58             } 59         }); 60  61     } 62  63 }

截取一段日志 可以得到我们注释里的结论：

 1 take-- PriorityEntity  [priority=8, index=0] left:-- [[]]  2 take-- PriorityEntity  [priority=1, index=1] left:-- [[]]  3 take-- PriorityEntity  [priority=8, index=2] left:-- [[]]  4 take-- PriorityEntity  [priority=7, index=3] left:-- [[PriorityEntity  [priority=8, index=4]]]  5 take-- PriorityEntity  [priority=8, index=4] left:-- [[PriorityEntity  [priority=9, index=5]]]  6 take-- PriorityEntity  [priority=1, index=6] left:-- [[PriorityEntity  [priority=8, index=7], PriorityEntity  [priority=9, index=5]]]  7 take-- PriorityEntity  [priority=8, index=7] left:-- [[PriorityEntity  [priority=9, index=5]]]  8 take-- PriorityEntity  [priority=2, index=8] left:-- [[PriorityEntity  [priority=9, index=5]]]  9 take-- PriorityEntity  [priority=9, index=5] left:-- [[]] 10 take-- PriorityEntity  [priority=5, index=9] left:-- [[]] 11 take-- PriorityEntity  [priority=4, index=10] left:-- [[]] 12 take-- PriorityEntity  [priority=4, index=13] left:-- [[PriorityEntity  [priority=6, index=11], PriorityEntity  [priority=6, index=12]]] 13 take-- PriorityEntity  [priority=3, index=14] left:-- [[PriorityEntity  [priority=6, index=16], PriorityEntity  [priority=6, index=12], PriorityEntity  [priority=6, index=11], PriorityEntity  [priority=8, index=15]]] 14 take-- PriorityEntity  [priority=6, index=16] left:-- [[PriorityEntity  [priority=6, index=12], PriorityEntity  [priority=8, index=15], PriorityEntity  [priority=6, index=11]]] 15 take-- PriorityEntity  [priority=6, index=12] left:-- [[PriorityEntity  [priority=6, index=17], PriorityEntity  [priority=8, index=15], PriorityEntity  [priority=6, index=11]]] 16 take-- PriorityEntity  [priority=6, index=17] left:-- [[PriorityEntity  [priority=6, index=11], PriorityEntity  [priority=8, index=15], PriorityEntity  [priority=8, index=18]]] 17 take-- PriorityEntity  [priority=6, index=11] left:-- [[PriorityEntity  [priority=8, index=18], PriorityEntity  [priority=8, index=15]]] 18 take-- PriorityEntity  [priority=4, index=19] left:-- [[PriorityEntity  [priority=8, index=18], PriorityEntity  [priority=8, index=15]]] 19 take-- PriorityEntity  [priority=8, index=18] left:-- [[PriorityEntity  [priority=8, index=15]]] 20 take-- PriorityEntity  [priority=7, index=20] left:-- [[PriorityEntity  [priority=8, index=15]]] 21 take-- PriorityEntity  [priority=2, index=21] left:-- [[PriorityEntity  [priority=4, index=22], PriorityEntity  [priority=8, index=15]]] 22 take-- PriorityEntity  [priority=4, index=22] left:-- [[PriorityEntity  [priority=8, index=23], PriorityEntity  [priority=8, index=15]]] 23 take-- PriorityEntity  [priority=8, index=23] left:-- [[PriorityEntity  [priority=8, index=15]]] 24 take-- PriorityEntity  [priority=5, index=24] left:-- [[PriorityEntity  [priority=8, index=15]]] 25 take-- PriorityEntity  [priority=2, index=25] left:-- [[PriorityEntity  [priority=8, index=26], PriorityEntity  [priority=8, index=15]]] 26 take-- PriorityEntity  [priority=3, index=27] left:-- [[PriorityEntity  [priority=4, index=28], PriorityEntity  [priority=8, index=15], PriorityEntity  [priority=8, index=26]]] 27 take-- PriorityEntity  [priority=1, index=30] left:-- [[PriorityEntity  [priority=4, index=28], PriorityEntity  [priority=7, index=29], PriorityEntity  [priority=8, index=26], PriorityEntity  [priority=8, index=15], PriorityEntity  [priority=8, index=31]]] 28 take-- PriorityEntity  [priority=4, index=28] left:-- [[PriorityEntity  [priority=7, index=29], PriorityEntity  [priority=8, index=15], PriorityEntity  [priority=8, index=26], PriorityEntity  [priority=9, index=32], PriorityEntity  [priority=8, index=31]]]

有兴趣的话可以看看java里面 有几种类 都实现了AbstractQueue,可以挑选出适合自己业务里的队列，减少开发难度

1 public abstract class AbstractQueue<E> 2     extends AbstractCollection<E> 3     implements Queue<E> {

4.CopyOnWriteArrayList

考虑这样一种场景，一个list，被好几个线程同时读写，那一般都会报错。

 1 Exception in thread "pool-1-thread-7" java.util.ConcurrentModificationException  2     at java.util.ArrayList$Itr.checkForComodification(Unknown Source)  3     at java.util.ArrayList$Itr.next(Unknown Source)  4     at com.android.testclass.Test7$ReadTask.run(Test7.java:35)  5     at java.util.concurrent.ThreadPoolExecutor.runWorker(Unknown Source)  6     at java.util.concurrent.ThreadPoolExecutor$Worker.run(Unknown Source)  7     at java.lang.Thread.run(Unknown Source)  8 Exception in thread "pool-1-thread-6" java.util.ConcurrentModificationException  9     at java.util.ArrayList$Itr.checkForComodification(Unknown Source) 10     at java.util.ArrayList$Itr.next(Unknown Source) 11     at com.android.testclass.Test7$ReadTask.run(Test7.java:35) 12     at java.util.concurrent.ThreadPoolExecutor.runWorker(Unknown Source) 13     at java.util.concurrent.ThreadPoolExecutor$Worker.run(Unknown Source) 14     at java.lang.Thread.run(Unknown Source)

于是很多人就喜欢用Collections.synchronizedList() 来处理，但是这样做在很多时候效率是低的，比如

假设现在告诉你，你需要设计一个缓存list，你就应该使用CopyOnWrite这个类了，因为缓存大家都知道，读操作比较多，而写操作除了在初始建立缓存的阶段，其他时候很少使用。

他的原理也很简单，就是你在用迭代器写操作的时候 是把原来的数据拷贝了一份镜像在内存中，而你在读的时候 是读的本体，写操作写完以后才会覆盖掉原来的本地。所以可以

得知 这个类对于频繁读的同步性list 是非常有效的。使用方法也很简单。

1         List<String> list = new CopyOnWriteArrayList<String>();

5.ThreadLocal

这个类也是很有效，很多开源作者喜欢用的一个类，他主要的作用是为每个线程创造一个变量的副本互相不会影响。很多人不理解这句话，

对于多线程操作来说 分为两种

1 第一种，线程和线程之间互相读取操作，比如全局的计数器这种，a线程要加，b线程也要加，每次加的时候 都要读取最新的计数器的状态。这是最常见的一种同步操作。

2 第二种，session，session一个用户一个，互相不影响，大家维持自己的就可以，他的目标就是a的seesion a自己操作 保存 读取，b的seesion也是自己维护，和其他人无关。

换一句话说 如果你需要多个线程之间通信，那就用同步机制，

如果你不需要线程与线程之间通信，只要互相别影响 不让他们发生冲突 则threadlocal是最佳选择。

 1 package com.android.testclass;  2   3 public class Test8 {  4   5     static final ThreadLocal<Integer> local = new ThreadLocal<Integer>() {  6   7         protected Integer initialValue() {  8   9             return 0; 10         }; 11  12     }; 13  14     public static void main(String[] args) { 15         // TODO Auto-generated method stub 16  17         Thread[] threads = new Thread[5]; 18         for (int i = 0; i < 5; i++) { 19             threads[i] = new Thread(new Runnable() { 20  21                 @Override 22                 public void run() { 23                     // TODO Auto-generated method stub 24  25                     int num = local.get(); 26                     for (int i = 0; i < 5; i++) { 27                         num++; 28                     } 29                     local.set(num); 30                     System.out.println(Thread.currentThread().getName() + " : " + local.get()); 31  32                 } 33             }, "thread-" + i); 34         } 35  36         for (Thread thread : threads) { 37             thread.start(); 38         } 39  40     } 41  42 }

看下输出

1 thread-0 : 5 2 thread-4 : 5 3 thread-1 : 5 4 thread-3 : 5 5 thread-2 : 5

接着看下面的

 1 package com.android.testclass;  2   3 public class Test9 {  4   5     private static Index num = new Index();  6     // 创建一个Index类型的本地变量  7     private static ThreadLocal<Index> local = new ThreadLocal<Index>() {  8         @Override  9         protected Index initialValue() { 10             return num; 11         } 12     }; 13  14     public static void main(String[] args) throws InterruptedException { 15         Thread[] threads = new Thread[5]; 16         for (int j = 0; j < 5; j++) { 17             threads[j] = new Thread(new Runnable() { 18                 @Override 19                 public void run() { 20                     // 取出当前线程的本地变量，并累加1000次 21                     Index index = local.get(); 22                     for (int i = 0; i < 1000; i++) { 23                         index.increase(); 24                     } 25                     System.out.println(Thread.currentThread().getName() + " : " + index.num); 26  27                 } 28             }, "Thread-" + j); 29         } 30         for (Thread thread : threads) { 31             thread.start(); 32         } 33     } 34  35     static class Index { 36         int num; 37  38         public void increase() { 39             num++; 40         } 41     } 42  43 }

看输出

Thread-1 : 2594 Thread-4 : 3594 Thread-2 : 2594 Thread-0 : 2594 Thread-3 : 4594

是因为第10行，那边放的是一个静态变量的引用，所以输出的结果不是我们想象的

其实只要改成

1 private static ThreadLocal<Index> local = new ThreadLocal<Index>() { 2         @Override 3         protected Index initialValue() { 4             return new Index(); 5         } 6     };

结果就是正确的：

1 Thread-2 : 1000 2 Thread-3 : 1000 3 Thread-0 : 1000 4 Thread-4 : 1000 5 Thread-1 : 1000