使用多线程的数独求解器

时间:2018-02-20 10:09:17

标签: java multithreading performance

我使用回溯实现了一个数独求解器但是我遇到了性能问题,主要问题是isAvailable()函数检查该数字是否对当前位置有效。不使用线程的执行时间需要61ms

protected boolean flag;

public boolean isAvailable(int sudoku[][], int row, int col, int num){
    flag = true;

    int rowStart = (row / 3) * 3;
    int colStart = (col / 3) * 3;

    for (int i = 0; i < 9; ++i) {
        if (sudoku[row][i] == num) {
            flag = false;
        }
    }

    for (int i = 0; i < 9; ++i) {
        if (sudoku[i][col] == num) {
            flag = false;
        }
    }

    for (int i = rowStart; i < (rowStart + 3); ++i) {
        for (int j = colStart; j < (colStart + 3); ++j) {
            if (sudoku[i][j] == num)
                flag = false;
        }
    }

    return flag;
}

结果:

Answer found
Time elapsed = 61.63ms
8 1 2 7 5 3 6 4 9 
9 4 3 6 8 2 1 7 5 
6 7 5 4 9 1 2 8 3 
1 5 4 2 3 7 8 9 6 
3 6 9 8 4 5 7 2 1 
2 8 7 1 6 9 5 3 4 
5 2 1 9 7 4 3 6 8 
4 3 8 5 2 6 9 1 7 
7 9 6 3 1 8 4 5 2 

但是如果我使用不同的线程一个用于检查另一个用于检查列而另一个用于检查列,我预计执行时间会缩短,但它花费了大约312s

protected boolean flag;

public boolean isAvailable(int sudoku[][], int row, int col, int num) throws InterruptedException {
    flag = true;

    int rowStart = (row / 3) * 3;
    int colStart = (col / 3) * 3;

    Thread t1 = new Thread(new Runnable() {
        public void run() {
            for (int i = 0; i < 9; ++i) {
                if (sudoku[row][i] == num) {
                    flag = false;
                }
            }
        }

    });
    t1.start();

    Thread t2 = new Thread(new Runnable() {
        public void run() {
            for (int i = 0; i < 9; ++i) {
                if (sudoku[i][col] == num) {
                    flag = false;
                }
            }
        }

    });
    t2.start();


    Thread t3 = new Thread(new Runnable() {
        public void run() {
            for (int i = rowStart; i < (rowStart + 3); ++i) {
                for (int j = colStart; j < (colStart + 3); ++j) {
                    if (sudoku[i][j] == num)
                        flag = false;
                }
            }
        }
    });
    t3.start();

    t1.join();
    t2.join();
    t3.join();

    return flag;
}

结果:

Answer found
Time elapsed = 312514.63ms
8 1 2 7 5 3 6 4 9 
9 4 3 6 8 2 1 7 5 
6 7 5 4 9 1 2 8 3 
1 5 4 2 3 7 8 9 6 
3 6 9 8 4 5 7 2 1 
2 8 7 1 6 9 5 3 4 
5 2 1 9 7 4 3 6 8 
4 3 8 5 2 6 9 1 7 
7 9 6 3 1 8 4 5 2 

如果您需要我的完整代码:

import java.text.DecimalFormat;

public class Main {

    public static void main(String[] args){

          int sudoku[][] = {{8, 0, 0, 0, 0, 0, 0, 0, 0},
                            {0, 0, 3, 6, 0, 0, 0, 0, 0},
                            {0, 7, 0, 0, 9, 0, 2, 0, 0},
                            {0, 5, 0, 0, 0, 7, 0, 0, 0},
                            {0, 0, 0, 0, 4, 5, 7, 0, 0},
                            {0, 0, 0, 1, 0, 0, 0, 3, 0},
                            {0, 0, 1, 0, 0, 0, 0, 6, 8},
                            {0, 0, 8, 5, 0, 0, 0, 1, 0},
                            {0, 9, 0, 0, 0, 0, 4, 0, 0}};

        Main m = new Main();
        long starttime = System.nanoTime();
        boolean k = m.fillsudoku(sudoku, 0, 0);
        long endtime = System.nanoTime();

        DecimalFormat df = new DecimalFormat("#.##");

        if (k) {
            System.out.println("Answer found");
            System.out.println("Time elapsed = " + df.format((double) (endtime - starttime) / 1000000) + "ms");
            for (int i = 0; i < 9; i++) {
                for (int j = 0; j < 9; j++) {
                    System.out.print(sudoku[i][j] + " ");
                }
                System.out.println();
            }
        } else {
            System.out.println("Not found");
        }
    }

    public boolean fillsudoku(int sudoku[][], int row, int col){
        if(row < 9){
            if(sudoku[row][col] != 0){
                if(col < 8){
                    return fillsudoku(sudoku, row, col+1);
                }
                else if(row < 8){
                    return fillsudoku(sudoku, row+1, 0);
                }
                return true;
            }
            else{
                for(int i=1;i<=9;i++){
                    if(isAvailable(sudoku, row, col, i)){        // <- checking function
                        sudoku[row][col] = i;
                        if(col == 8){
                            if(fillsudoku(sudoku, row+1, 0)){
                                return true;
                            }
                        }
                        else{
                            if(fillsudoku(sudoku, row, col+1)){
                                return true;
                            }
                        }
                        sudoku[row][col] = 0;
                    }
                }
                return false;
            }
        }
        return true;
    }
    protected boolean flag;

    public boolean isAvailable(int sudoku[][], int row, int col, int num){
    ........
    ........
    }
}

如何使用线程,任何建议来提高性能,我是否正确思考。

2 个答案:

答案 0 :(得分:4)

不幸的是,你在这里没有从多线程中获得任何好处 每次方法调用创建3个线程比连续运行3个简单循环更昂贵 实际上,27个检查和重新分配迭代不值得并行化。

说到改进,我们可以通过使用ExecutorService的线程池来重写方法。我试过了,结果如下:

54.25ms     // your plain version
65397.06ms  // your multithreading version
5500.82ms   // ExecutorService version

我只是将Thread变量转换为Runnable个变量,将它们传递给ExecutorService#submit并调用Future#get

Runnable r1 = () -> { ... };
// other Runnables

final Future<?> f1 = executorService.submit(r1);
// other Futures

try {
    f1.get();
    // other gets
} catch (ExecutionException e) { ... }

答案 1 :(得分:1)

你可以有效地使用线程,但不能让他们做这么小的工作。您正在为单个字段顺序尝试多个i,这是并行化的最佳位置。请注意,您必须深度克隆sudoku数组,以便您的线程不会相互踩踏。通过这种方式可以轻松实现接近CPU数量的加速。