代码之家 › 专栏 › 技术社区 › Hleb

为什么使用ForkJoin框架的阶乘计算比使用递归的顺序计算慢

forkjoinpool fork-join factorial java

Hleb · 技术社区 · 7 年前

我用两种方法实现了阶乘计算——使用递归和使用ForkJoin框架。我原以为ForkJoin实现会更快,但在我尝试测试的所有情况下,它总是较慢(我尝试为不同的数字计算阶乘,还尝试调整并行度)。以下是我的两个实现—没有FJ:

public class SequentialFactorial {

    public static void main(String[] args) {
        long start = System.nanoTime();
        System.out.println(calculateFactorial(BigInteger.valueOf(500)));
        long end = System.nanoTime();
        System.out.println((end - start) + " nanoseconds for sequential");
    }

    private static BigInteger calculateFactorial(BigInteger n) {
        if (n.compareTo(BigInteger.valueOf(2)) <= 0) {
            return n;
        }

        return n.multiply(calculateFactorial(n.subtract(BigInteger.ONE)));
    }
}

有了它:

public class ForkJoinFactorial {

    public static void main( String[] args ) {
        ForkJoinPool forkJoinPool = ForkJoinPool.commonPool();
        long start = System.nanoTime();
        System.out.println(forkJoinPool.invoke(new FactorialTask(BigInteger.valueOf(500))));
        long end = System.nanoTime();
        System.out.println((end - start) + " nanoseconds for FJP");
    }
}

public class FactorialTask extends RecursiveTask<BigInteger> {
    private BigInteger number;

    public FactorialTask(BigInteger number) {
        this.number = number;
    }

    @Override
    protected BigInteger compute() {
        if (number.compareTo(BigInteger.valueOf(1)) <= 0) {
            return MathUtil.calculateFactorial(number);
        }
        FactorialTask subtask = new FactorialTask(number.subtract(BigInteger.ONE));
        subtask.fork();
        return number.multiply(subtask.join());
    }
}

在我上次的测试中,我得到了这样的结果: