摘自马士兵java并发编程
一、认识Executor、ExecutorService、Callable、Executors
/** * 认识Executor */ package yxxy.c_026; import java.util.concurrent.Executor; public class T01_MyExecutor implements Executor { public static void main(String[] args) { new T01_MyExecutor().execute(new Runnable(){ @Override public void run() { System.out.println("hello executor"); } }); } @Override public void execute(Runnable command) { //new Thread(command).run(); command.run(); } }
Executor执行器是一个接口,只有一个方法execute执行任务,在java的线程池的框架里边,这个是最顶层的接口;
ExecutorService:从Executor接口继承。
Callable:里面call方法,和Runnable接口很像,设计出来都是被其他线程调用的;但是Runnable接口里面run方法是没有返回值的也不能抛出异常;而call方法有返回值可以抛异常;
Executors: 操作Executor的一个工具类;以及操作ExecutorService,ThreadFactory,Callable等;
二、ThreadPool:
/** * 线程池的概念 */ package yxxy.c_026; import java.util.concurrent.ExecutorService; import java.util.concurrent.Executors; import java.util.concurrent.TimeUnit; public class T05_ThreadPool { public static void main(String[] args) throws InterruptedException { ExecutorService service = Executors.newFixedThreadPool(5); //execute submit for (int i = 0; i < 6; i++) { service.execute(() -> { try { TimeUnit.MILLISECONDS.sleep(500); } catch (InterruptedException e) { e.printStackTrace(); } System.out.println(Thread.currentThread().getName()); }); } System.out.println(service); service.shutdown(); System.out.println(service.isTerminated()); System.out.println(service.isShutdown()); System.out.println(service); TimeUnit.SECONDS.sleep(5); System.out.println(service.isTerminated()); System.out.println(service.isShutdown()); System.out.println(service); } }
console:
java.util.concurrent.ThreadPoolExecutor@53d8d10a[Running, pool size = 5, active threads = 5, queued tasks = 1, completed tasks = 0] false true java.util.concurrent.ThreadPoolExecutor@53d8d10a[Shutting down, pool size = 5, active threads = 5, queued tasks = 1, completed tasks = 0] pool-1-thread-1 pool-1-thread-3 pool-1-thread-2 pool-1-thread-5 pool-1-thread-4 pool-1-thread-1 true true java.util.concurrent.ThreadPoolExecutor@53d8d10a[Terminated, pool size = 0, active threads = 0, queued tasks = 0, completed tasks = 6]
创建了一个线程池,扔了5个线程,接下来要执行6个任务,扔进去线程池里面就启一个线程帮你执行一个,因为这里最多就起5个线程,接下来扔第6个任务的时候,不好意思,它排队了,排在线程池所维护的一个任务队列里面,任务队列大多数使用的都是BlockingQueue,这是线程池的概念;
有什么好处?好处在于如果这个任务执行完了,这个线程不会消失,它执行完任务空闲下来了,如果有新的任务来的时候,直接交给这个线程来运行就行了,不需要新启动线程;从这个概念上讲,如果你的任务和线程池线程数量控制的比较好的情况下,你不需要启动新的线程就能执行很多很多的任务,效率会比较高,并发性好;
service.shutdown():关闭线程池,shutdown是正常的关闭,它会等所有的任务都执行完才会关闭掉;还有一个是shutdownNow,二话不说直接就给关了,不管线程有没有执行完;
service.isTerminated(): 代表的是这里所有执行的任务是不是都执行完了。isShutdown()为true,注意它关了但并不代表它执行完了,只是代表正在关闭的过程之中(注意打印Shutting down)
打印5个线程名字,而且第一个线程执行完了之后,第6个任务来了,第1个线程继续执行,不会有线程6;
当所有线程全部执行完毕之后,线程池的状态为Terminated,表示正常结束,complete tasks=6
线程池里面维护了很多线程,等着你往里扔任务,而扔任务的时候它可以维护着一个任务列表,还没有被执行的任务列表,同样的它还维护着另外一个队列,complete tasks,结束的任务队列,任务执行结束扔到这个队列里,所以,一个线程池维护着两个队列;
三、Future
/** * 认识future */ package yxxy.c_026; import java.util.concurrent.ExecutionException; import java.util.concurrent.ExecutorService; import java.util.concurrent.Executors; import java.util.concurrent.Future; import java.util.concurrent.FutureTask; import java.util.concurrent.TimeUnit; public class T06_Future { public static void main(String[] args) throws InterruptedException, ExecutionException { /*FutureTask<Integer> task = new FutureTask<Integer>(new Callable<Integer>(){ @Override public Integer call() throws Exception { TimeUnit.MILLISECONDS.sleep(3000); return 1000; } });*/ FutureTask<Integer> task = new FutureTask<>(()->{ TimeUnit.MILLISECONDS.sleep(3000); return 1000; }); new Thread(task).start(); System.out.println(task.get()); //阻塞 //******************************* ExecutorService service = Executors.newFixedThreadPool(5); Future<Integer> f = service.submit(()->{ TimeUnit.MILLISECONDS.sleep(5000); return 1; }); System.out.println(f.isDone()); System.out.println(f.get()); System.out.println(f.isDone()); } }
1000 false 1 true
Future: ExecutorService里面有submit方法,它的返回值是Future类型,因为你扔一个任务进去需要执行一段时间,未来的某一个时间点上,任务执行完了产生给你一个结果,这个Future代表的就是那个Callable的返回值;
四、并行计算的例子:
/** * 线程池的概念 * nasa */ package yxxy.c_026; import java.util.ArrayList; import java.util.List; import java.util.concurrent.Callable; import java.util.concurrent.ExecutionException; import java.util.concurrent.ExecutorService; import java.util.concurrent.Executors; import java.util.concurrent.Future; public class T07_ParallelComputing { public static void main(String[] args) throws InterruptedException, ExecutionException { long start = System.currentTimeMillis(); List<Integer> results = getPrime(1, 200000); long end = System.currentTimeMillis(); System.out.println(end - start); final int cpuCoreNum = 4; ExecutorService service = Executors.newFixedThreadPool(cpuCoreNum); MyTask t1 = new MyTask(1, 80000); //1-5 5-10 10-15 15-20 MyTask t2 = new MyTask(80001, 130000); MyTask t3 = new MyTask(130001, 170000); MyTask t4 = new MyTask(170001, 200000); Future<List<Integer>> f1 = service.submit(t1); Future<List<Integer>> f2 = service.submit(t2); Future<List<Integer>> f3 = service.submit(t3); Future<List<Integer>> f4 = service.submit(t4); start = System.currentTimeMillis(); f1.get(); f2.get(); f3.get(); f4.get(); end = System.currentTimeMillis(); System.out.println(end - start); } static class MyTask implements Callable<List<Integer>> { int startPos, endPos; MyTask(int s, int e) { this.startPos = s; this.endPos = e; } @Override public List<Integer> call() throws Exception { List<Integer> r = getPrime(startPos, endPos); return r; } } //判断是否是质数 static boolean isPrime(int num) { for(int i=2; i<=num/2; i++) { if(num % i == 0) return false; } return true; } static List<Integer> getPrime(int start, int end) { List<Integer> results = new ArrayList<>(); for(int i=start; i<=end; i++) { if(isPrime(i)) results.add(i); } return results; } }
console:
2280 818
第二种方式使用了一个线程池,一般线程池有多少个线程,数量多少合适是需要调整的,大多数情况下cpu有几个核至少就应该起多少个线程,可以多起一个但不能少于cpu核数,将20万分成了4段;
这里为什么不将20万平均分呢?
五、CachedThreadPool
package yxxy.c_026; import java.util.concurrent.ExecutorService; import java.util.concurrent.Executors; import java.util.concurrent.TimeUnit; public class T08_CachedPool { public static void main(String[] args) throws InterruptedException { ExecutorService service = Executors.newCachedThreadPool(); System.out.println(service); for (int i = 0; i < 2; i++) { service.execute(() -> { try { TimeUnit.MILLISECONDS.sleep(500); } catch (InterruptedException e) { e.printStackTrace(); } System.out.println(Thread.currentThread().getName()); }); } System.out.println(service); TimeUnit.SECONDS.sleep(80); //cachedthreadPool里面的线程空闲状态默认60s后销毁,这里保险起见 System.out.println(service); } }
console:
java.util.concurrent.ThreadPoolExecutor@7852e922[Running, pool size = 0, active threads = 0, queued tasks = 0, completed tasks = 0] java.util.concurrent.ThreadPoolExecutor@7852e922[Running, pool size = 2, active threads = 2, queued tasks = 0, completed tasks = 0] pool-1-thread-2 pool-1-thread-1 java.util.concurrent.ThreadPoolExecutor@7852e922[Running, pool size = 0, active threads = 0, queued tasks = 0, completed tasks = 2]
FixedThreadPool为固定个数的线程池;
CachedThreadPool:刚开始一个线程都没有,来一个任务就起一个线程,假设起了两个线程A,B,如果来了第三个任务,这时候恰好线程B任务执行完了,线程池里面有空闲的,这时候直接让线程池里空闲的线程B来执行;最多起多少个线程?你的系统能支撑多少个为止;默认的情况下,只要一个线程空闲的状态超过60s,这个线程就自动的销毁了,alivetime=60s;这个值也可以自己指定。
六、SingleThreadExecutor
package yxxy.c_026; import java.util.concurrent.ExecutorService; import java.util.concurrent.Executors; public class T09_SingleThreadPool { public static void main(String[] args) { ExecutorService service = Executors.newSingleThreadExecutor(); for(int i=0; i<5; i++) { final int j = i; service.execute(()->{ System.out.println(j + " " + Thread.currentThread().getName()); }); } } }
console:
0 pool-1-thread-1 1 pool-1-thread-1 2 pool-1-thread-1 3 pool-1-thread-1 4 pool-1-thread-1
SingleThreadExecutor:线程池里就1个线程;扔5个任务,也永远只有1个线程执行;
它能保证任务前后一定是顺序执行,先扔的任务一定先执行完;只有等第一个任务执行完才执行第二个任务。
七、ScheduledThreadPool
package yxxy.c_026; import java.util.Random; import java.util.concurrent.Executors; import java.util.concurrent.ScheduledExecutorService; import java.util.concurrent.TimeUnit; public class T10_ScheduledPool { public static void main(String[] args) { ScheduledExecutorService service = Executors.newScheduledThreadPool(4); service.scheduleAtFixedRate(()->{ try { TimeUnit.MILLISECONDS.sleep(new Random().nextInt(1000)); } catch (InterruptedException e) { e.printStackTrace(); } System.out.println(Thread.currentThread().getName()); }, 0, 500, TimeUnit.MILLISECONDS); } }
ScheduledThreadPool: 执行定时的任务,定时器线程池,一般可以用来替代timer,而且它里面的线程是可以复用的,第一个线程执行完了之后,任务来了如果第一个线程是空闲的,还可以拿第一个线程来执行。而Timer每次都是new一个新的线程。
scheduleAtFixedRate(
Runnable command, long initialDelay, long period,
TimeUnit unit),第1个参数是任务,第1个任务马上执行,每隔500毫秒这个任务重复执行。
八、WorkStealingPool
/** * */ package yxxy.c_026; import java.io.IOException; import java.util.concurrent.ExecutorService; import java.util.concurrent.Executors; import java.util.concurrent.TimeUnit; public class T11_WorkStealingPool { public static void main(String[] args) throws IOException { ExecutorService service = Executors.newWorkStealingPool(); int count = Runtime.getRuntime().availableProcessors(); //看cpu多少核的;如果是4核,默认就帮你起4个线程 System.out.println(count); service.execute(new R(1000)); for(int i=0; i<count; i++){ service.execute(new R(2000)); } //由于产生的是精灵线程(守护线程、后台线程),主线程不阻塞的话,看不到输出 System.in.read(); } static class R implements Runnable { int time; R(int t) { this.time = t; } @Override public void run() { try { TimeUnit.MILLISECONDS.sleep(time); } catch (InterruptedException e) { e.printStackTrace(); } System.out.println(time + " " + Thread.currentThread().getName()); } } }
console:
8 1000 ForkJoinPool-1-worker-1 2000 ForkJoinPool-1-worker-2 2000 ForkJoinPool-1-worker-0 2000 ForkJoinPool-1-worker-5 2000 ForkJoinPool-1-worker-3 2000 ForkJoinPool-1-worker-6 2000 ForkJoinPool-1-worker-7 2000 ForkJoinPool-1-worker-4 2000 ForkJoinPool-1-worker-1
WorkStealingPool:工作窃取,假设有3个线程A、B、C在运行,workStealing可以简单这么认为,每个线程都维护自己的一个队列,线程A的队列里头积累了5个任务,线程B的队列里1个任务,C的队列里2个任务;那么当线程B执行完任务之后,他会去别的线程池所维护的队列里面把任务偷过来继续执行,主动的找活干。
本质上是使用ForkJoinPool来实现的:
public static ExecutorService newWorkStealingPool() { return new ForkJoinPool (Runtime.getRuntime().availableProcessors(), ForkJoinPool.defaultForkJoinWorkerThreadFactory, null, true); }
例子解释:cpu多少核默认的起多少个线程,(这里是8),前面几个任务都扔给1-8个线程了,第9个任务来的时候在那里等着了,谁会去执行它呢?先执行完任务的这个线程会去执行。第1个线程只睡1s钟,首先执行完,所以第9个任务一定是第一个线程1去运行它,他会主动的把任务拿过去运行。
workStealing的线程是精灵线程,daemon线程,特点就是主线程main方法一旦结束了,它后台可能还在运行,但是你是看不到它任务输出的;这里Syetem.in.read()让主函数阻塞才能看到输出。debug的时候能看到Daemon Thread[ForkJoinPool-1-worker-1]。为什么用精灵线程?它是在后台不断的运行的,只要虚拟机不退出,这个线程就不会退出,你有任务来了之后,它永远会主动去拿。
workStealing用于什么场景:就说任务分配的不是很均匀,有的线程维护的任务队列比较长,有些线程执行完任务就结束了不太合适,所以他执行完了之后可以去别的线程维护的队列里去偷任务,这样效率更高。
九、ForkJoinPool
ForkJoinPool: forkjoin的意思就是如果有一个难以完成的大任务,需要计算量特别大,时间特别长,可以把大任务切分成一个个小任务,如果小任务还是太大,它还可以继续分,至于分成多少你可以自己指定,... 分完之后,把结果进行合并,最后合并到一起join一起,产生一个总的结果。而里面任务的切分你可以自己指定,线程的启动根据你任务切分的规则,由ForkJoinPool这个线程池自己来维护。
例子1:
package yxxy.c_026; import java.io.IOException; import java.util.Arrays; import java.util.Random; import java.util.concurrent.ForkJoinPool; import java.util.concurrent.RecursiveAction; import java.util.concurrent.RecursiveTask; public class T12_ForkJoinPool { static int[] nums = new int[1000000]; static final int MAX_NUM = 50000; static Random r = new Random(); static { for(int i=0; i<nums.length; i++) { nums[i] = r.nextInt(100); } System.out.println(Arrays.stream(nums).sum()); //stream api } static class AddTask extends RecursiveAction { int start, end; AddTask(int s, int e) { start = s; end = e; } @Override protected void compute() { if(end-start <= MAX_NUM) { long sum = 0L; for(int i=start; i<end; i++) sum += nums[i]; System.out.println("from:" + start + " to:" + end + " = " + sum); } else { int middle = start + (end-start)/2; AddTask subTask1 = new AddTask(start, middle); AddTask subTask2 = new AddTask(middle, end); subTask1.fork(); subTask2.fork(); } } } public static void main(String[] args) throws IOException { ForkJoinPool fjp = new ForkJoinPool(); AddTask task = new AddTask(0, nums.length); fjp.execute(task); System.in.read(); } }
console:
49494882 from:906250 to:937500 = 1545274 from:968750 to:1000000 = 1537201 from:593750 to:625000 = 1548289 from:718750 to:750000 = 1546396 from:468750 to:500000 = 1550373 from:843750 to:875000 = 1543421 from:218750 to:250000 = 1549856 from:93750 to:125000 = 1548384 from:562500 to:593750 = 1541814 from:812500 to:843750 = 1547885 from:187500 to:218750 = 1546831 from:687500 to:718750 = 1554064 from:437500 to:468750 = 1547434 from:937500 to:968750 = 1547676 from:875000 to:906250 = 1551839 from:62500 to:93750 = 1548576 from:531250 to:562500 = 1550943 from:656250 to:687500 = 1544991 from:156250 to:187500 = 1548367 from:406250 to:437500 = 1539881 from:125000 to:156250 = 1548128 from:500000 to:531250 = 1545229 from:781250 to:812500 = 1544296 from:625000 to:656250 = 1545283 from:375000 to:406250 = 1553931 from:31250 to:62500 = 1544024 from:750000 to:781250 = 1543573 from:343750 to:375000 = 1546407 from:0 to:31250 = 1539743 from:281250 to:312500 = 1549470 from:312500 to:343750 = 1552190 from:250000 to:281250 = 1543113
例子解释:
对数组中100万个数求和计算,第一种方式是普通的将所有数加在一起(for循环);
第二种方式使用ForkJoinPool计算,分而治之,它里面执行的任务必须是ForkJoinTask,这个任务可以自动进行切分,一般用的时候从RecursiveAction或RecursiveTask继承,RecursiveTask递归任务,因为它切分任务还可以在切分。RecursiveAction没有返回值,RecursiveTask有返回值。
例子2:
package yxxy.c_026; import java.io.IOException; import java.util.Arrays; import java.util.Random; import java.util.concurrent.ForkJoinPool; import java.util.concurrent.RecursiveAction; import java.util.concurrent.RecursiveTask; public class T12_ForkJoinPool { static int[] nums = new int[1000000]; static final int MAX_NUM = 50000; static Random r = new Random(); static { for(int i=0; i<nums.length; i++) { nums[i] = r.nextInt(100); } System.out.println(Arrays.stream(nums).sum()); //stream api } static class AddTask extends RecursiveTask<Long> { int start, end; AddTask(int s, int e) { start = s; end = e; } @Override protected Long compute() { if(end-start <= MAX_NUM) { long sum = 0L; for(int i=start; i<end; i++) sum += nums[i]; return sum; } int middle = start + (end-start)/2; AddTask subTask1 = new AddTask(start, middle); AddTask subTask2 = new AddTask(middle, end); subTask1.fork(); subTask2.fork(); return subTask1.join() + subTask2.join(); } } public static void main(String[] args) throws IOException { ForkJoinPool fjp = new ForkJoinPool(); AddTask task = new AddTask(0, nums.length); fjp.execute(task); long result = task.join(); System.out.println(result); } }
console:
49498457 49498457
和例子1差不多,唯一的区别是有返回值了,RecursiveTask<V>中的V泛型就是返回值类型。
long result = task.join(),因为join本身就是阻塞的,只有等所有的都执行完了,最后才得出总的执行结果。所以不需要System.in.read了;
十、自定义线程池 ThreadPoolExecutor
ThreadPoolExecutor:大多数的线程池的实现背后调用的都是ThreadPoolExecutor(前面6种就ForkJoinPool不是),它是线程池通用的一个类,可以自定义线程池;
ThreadPoolExecutor(int corePoolSize, int maximumPoolSize, long keepAliveTime, TimeUnit unit, BlockingQueue<Runnable> workQueue, ThreadFactory threadFactory, RejectedExecutionHandler handler)
corePoolSize:核心的线程池里的线程数,自己指定。
maximumPoolSize:最多这个线程池里装多少个线程;
keepAliveTime:线程呆多久没有任务传给它就会消失;
unit:和上面统一指定的;
blockingQueue:真正的装任务的容器,往往都是用blockingQueue;阻塞式的;任务来了就扔进去,什么时候用到了都可以取。
例如,fixedThreadPool的实现是:
public static ExecutorService newFixedThreadPool(int nThreads) { return new ThreadPoolExecutor(nThreads, nThreads, 0L, TimeUnit.MILLISECONDS, new LinkedBlockingQueue<Runnable>()); }
十一、parallel stream
package yxxy.c_026; import java.util.ArrayList; import java.util.List; import java.util.Random; public class T14_ParallelStreamAPI { public static void main(String[] args) { List<Integer> nums = new ArrayList<>(); Random r = new Random(); for(int i=0; i<10000; i++) nums.add(1000000 + r.nextInt(1000000)); //System.out.println(nums); long start = System.currentTimeMillis(); nums.forEach(v->isPrime(v)); long end = System.currentTimeMillis(); System.out.println(end - start); //使用parallel stream api start = System.currentTimeMillis(); nums.parallelStream().forEach(T14_ParallelStreamAPI::isPrime); end = System.currentTimeMillis(); System.out.println(end - start); } static boolean isPrime(int num) { for(int i=2; i<=num/2; i++) { if(num % i == 0) return false; } return true; } }
console:
1526 337
paralleStream(): 运用多线程,如果把这1万个数看成是数据流,我们用多线程去访问里面的数,共同来做计算,默认使用多线程。
---------------