cache友好代码--以矩阵乘法为例

#include <iostream>
#include <ctime>
#define N 1024
using namespace std;
int A[N][N];
int B[N][N];
int C2[N][N];
int C3[N][N];
int C1[N][N];
void Random(int(*a)[N], int n, int l, int r)//生成范围在l~r的随机数 
{
	srand(time(0));  //设置时间种子
	for (int i = 0; i < n; ++i) {
		for (int j = 0; j < n; ++j) {
			a[i][j] = rand() % (r - l + 1) + l;//生成区间r~l的随机数 
		}
	}
}

int main()
{
		Random(A, N, 0, 10);
		Random(B, N, 0, 10);
		clock_t start, end;
		// cache不友好方式
		start = clock();
		for (int i = 0; i < N; ++i) {
			for (int j = 0; j < N; ++j) {
				int c = 0;
				for (int k = 0; k < N; ++k) {
					c += A[i][k] * B[k][j];
				}
				C1[i][j] = c;
			}
		}
		end = clock();
		cout << "不友好方式" << (double)((double)end - start) / CLOCKS_PER_SEC << endl;
		// cache 友好
		for (int i = 0; i < N; ++i) {
			for (int j = 0; j < N; ++j) {
				C2[i][j] = 0;
			}
		}
		start = clock();
		for (int i = 0; i < N; ++i) {
			for (int k = 0; k < N; ++k) {
				for (int j = 0; j < N; ++j) {
					C2[i][j] += A[i][k] * B[k][j];
				}
			}
		}
		end = clock();
		cout << "友好方式1" << (double)((double)end - start) / CLOCKS_PER_SEC << endl;

		// cache 友好
		for (int i = 0; i < N; ++i) {
			for (int j = 0; j < N; ++j) {
				C3[i][j] = 0;
			}
		}
		start = clock();
		for (int k = 0; k < N; ++k) {
			for (int i = 0; i < N; ++i) {
				for (int j = 0; j < N; ++j) {
					C3[i][j] += A[i][k] * B[k][j];
				}
			}
		}
		end = clock();
		cout << "不友好方式2" << (double)((double)end - start) / CLOCKS_PER_SEC << endl;
		for (int i = 0; i < N; ++i) {
			for (int j = 0; j < N; ++j) {
				if ((C1[i][j] != C2[i][j])|| (C1[i][j] != C3[i][j])) {
					cout << "calculate error!!!" << endl;
					cout << "(" << i << ", " << j << ") " <<
						"C1: " << C1[i][j] <<
						" <-> C2: " << C2[i][j] << endl;
					return 0;
				}
			}
		}
		cout << "success!!!" << endl;
}