矩阵乘法错误：分段错误（核心转储）

Question

I have the following code, and I want to calculate the running time of several matrix multiplications, with different sizes. I started with matrix size of 100, and moved till 500, but when I try 1000, I get an error saying: Segmentation fault (core dumped), so I assume it has to do something with the memory. I want to calculate the running time even of the matrices with sizes of 5000 and maybe 10000. Anyone can help me solve my problem?

#include <stdio.h>
#include <stdlib.h>
#include <time.h>
#define SIZE 1000

int main(void)
{
    int A[SIZE][SIZE], B[SIZE][SIZE], C[SIZE][SIZE] = {0};
    int i, j, k;

    srand(time(NULL));

    for(i = 0; i < SIZE; i++)
    {
        for(j = 0; j < SIZE; j++)
        {
            A[i][j] = rand()%100;
            B[i][j] = rand()%100;
        }
    }

    clock_t begin, end;
    double time_spent;

    begin = clock();

    for(i = 0; i < SIZE; i++)
        for(j = 0; j < SIZE; j++)
            for(k = 0; k < SIZE; k++)
                C[i][j] += A[i][k] * B[k][j];

    end = clock();

    time_spent = (double)(end - begin) / CLOCKS_PER_SEC;

    printf("Elapsed time: %.2lf seconds.\n", time_spent);

    return 0;
}

Answer 1

This is a classic stack overflow - you have 3 local arrays each of approx 4 MB in size (assuming 32 bit ints) on your stack for a total of around 12 MB. On most modern operating systems the stack is typically 8 MB or less. Either make the variables static, eg

static int A[SIZE][SIZE], B[SIZE][SIZE], C[SIZE][SIZE] = {0};

so that they no longer reside on the stack, or allocate them dynamically.