多線程計算卷积c++代碼

以下是一個基本的多線程計算卷積的代碼，使用OpenMP庫實現。該代碼使用了一個矩陣乘法的方法來計算卷積，並且使用了多線程來加速計算。

#include <iostream>
#include <omp.h>

using namespace std;

void convolve(int* input, int* kernel, int* output, int input_width, int input_height, int kernel_width, int kernel_height)
{
    int output_width = input_width - kernel_width + 1;
    int output_height = input_height - kernel_height + 1;

    #pragma omp parallel for
    for(int i = 0; i < output_width; i++)
    {
        for(int j = 0; j < output_height; j++)
        {
            int sum = 0;
            for(int k = 0; k < kernel_width; k++)
            {
                for(int l = 0; l < kernel_height; l++)
                {
                    int input_index = (i + k) * input_height + (j + l);
                    int kernel_index = k * kernel_height + l;
                    sum += input[input_index] * kernel[kernel_index];
                }
            }
            int output_index = i * output_height + j;
            output[output_index] = sum;
        }
    }
}

int main()
{
    // Initialize input and kernel matrices
    int input_width = 10;
    int input_height = 10;
    int kernel_width = 3;
    int kernel_height = 3;

    int input[input_width * input_height];
    int kernel[kernel_width * kernel_height];
    int output[(input_width - kernel_width + 1) * (input_height - kernel_height + 1)];

    for(int i = 0; i < input_width * input_height; i++)
    {
        input[i] = rand() % 10;
    }

    for(int i = 0; i < kernel_width * kernel_height; i++)
    {
        kernel[i] = rand() % 10;
    }

    // Calculate convolution using multiple threads
    double start_time = omp_get_wtime();
    convolve(input, kernel, output, input_width, input_height, kernel_width, kernel_height);
    double end_time = omp_get_wtime();

    // Print output
    cout << "Input Matrix:" << endl;
    for(int i = 0; i < input_width; i++)
    {
        for(int j = 0; j < input_height; j++)
        {
            cout << input[i * input_height + j] << "\t";
        }
        cout << endl;
    }

    cout << endl << "Kernel Matrix:" << endl;
    for(int i = 0; i < kernel_width; i++)
    {
        for(int j = 0; j < kernel_height; j++)
        {
            cout << kernel[i * kernel_height + j] << "\t";
        }
        cout << endl;
    }

    cout << endl << "Output Matrix:" << endl;
    for(int i = 0; i < (input_width - kernel_width + 1); i++)
    {
        for(int j = 0; j < (input_height - kernel_height + 1); j++)
        {
            cout << output[i * (input_height - kernel_height + 1) + j] << "\t";
        }
        cout << endl;
    }

    cout << endl << "Execution time: " << end_time - start_time << " seconds" << endl;

    return 0;
}