2021-10-25

Java 7 开始引入了一种新的 Fork/Join 线程池，它可以执行一种特殊的任务：把一个大任务拆成多个小任务并行执行。即“分而治之”。

ForkJoin 将相同的计算任务通过多线程的执行，提高计算速度。google 中的大数据处理框架MapReduce 就通过 ForkJoin 的思想，提高大数据的处理速度。

使用 ForkJoin 框架，需要创建一个 ForkJoin 的任务。因为 ForkJoin 框架为我们提供了无返回值的 RecursiveAction 和有返回值的 RecursiveTask。源码如下：

public abstract class RecursiveAction extends ForkJoinTask {
    private static final long serialVersionUID = 5232453952276485070L;

    
    protected abstract void compute();

    
    public final Void getRawResult() { return null; }

    
    protected final void setRawResult(Void mustBeNull) { }

    
    protected final boolean exec() {
        compute();
        return true;
    }

}

public abstract class RecursiveTask extends ForkJoinTask {
    private static final long serialVersionUID = 5232453952276485270L;

    
    V result;

    
    protected abstract V compute();

    public final V getRawResult() {
        return result;
    }

    protected final void setRawResult(V value) {
        result = value;
    }

    
    protected final boolean exec() {
        result = compute();
        return true;
    }

}

我们只需要继承JDK为我们提供的其中一个并且实现compute方法。例如，求和:

public class ForkJoinRecursiveAction {

    private final static int MAX_THRESHOLD = 3;

    private final static AtomicInteger SUM = new AtomicInteger(0);

    public static void main(String[] args) throws InterruptedException {

        final ForkJoinPool forkJoinPool = new ForkJoinPool();

        forkJoinPool.submit(new CalculateRecursiveAction(0, 10));

        forkJoinPool.awaitTermination(10, TimeUnit.SECONDS);

        Optional.of(SUM).ifPresent(System.out::println);

    }

    private static class CalculateRecursiveAction extends RecursiveAction {

        private final int start;

        private final int end;

        private CalculateRecursiveAction(int start, int end) {
            this.start = start;
            this.end = end;
        }

        @Override
        protected void compute() {

            if ((end - start) <= MAX_THRESHOLD) {
                SUM.addAndGet(IntStream.rangeClosed(start, end).sum());
            } else {
                int middle = (start + end) / 2;
                CalculateRecursiveAction leftAction = new CalculateRecursiveAction(start, middle);
                CalculateRecursiveAction rightAction = new CalculateRecursiveAction(middle + 1, end);
                leftAction.fork();
                rightAction.fork();
            }
        }
    }
}

public class ForkJoinRecursiveTask {

    private final static int MAX_THRESHOLD = 200;

    public static void main(String[] args) {
        final ForkJoinPool forkJoinPool = new ForkJoinPool();
        ForkJoinTask future = forkJoinPool.submit(new CalculatedRecursiveTask(0, 1000));
        try {
            Integer result = future.get();
            System.out.println(result);
        } catch (InterruptedException e) {
            e.printStackTrace();
        } catch (ExecutionException e) {
            e.printStackTrace();
        }
    }

    private static class CalculatedRecursiveTask extends RecursiveTask {

        private final int start;

        private final int end;

        CalculatedRecursiveTask(int start, int end) {
            this.start = start;
            this.end = end;
        }

        @Override
        protected Integer compute() {
            if (end - start <= MAX_THRESHOLD) {
                return IntStream.rangeClosed(start, end).sum();
            } else {
                int middle = (start + end) / 2;
                CalculatedRecursiveTask leftTask = new CalculatedRecursiveTask(start, middle);
                CalculatedRecursiveTask rightTask = new CalculatedRecursiveTask(middle + 1, end);

                leftTask.fork();
                rightTask.fork();

                return leftTask.join() + rightTask.join();
            }
        }
    }
}

task 要通过 ForkJoinPool 来执行，分割的子任务也会添加到当前工作线程的双端队列中，
进入队列的头部。当一个工作线程中没有任务时，会从其他工作线程的队列尾部获取一个任务(工作窃取)。