矩阵相乘哪里错了求大神！

[怪我_喽]

#include "device_launch_parameters.h"
#include<iostream>
#include<stdio.h>
#include "cuda_runtime.h"
#include "device_functions.h"

#define TILE_WIDTH 2
//定义一个4*4的矩阵
#define WIDTH 4
#define HEIGHT 4

__global__ void MatrixMulKernel(double * _A, double *_B, double * _C, int _wa, int _wb)
{
	
	//找出该线程所在的行列

	int row = blockIdx.y*blockDim.y + threadIdx.y;//该矩阵的行索引
	int col = blockIdx.x*blockDim.x + threadIdx.x;//该矩阵的类索引。。。
	//int row = threadIdx.y;//当用一个block的时候
	//int col = threadIdx.x;
	//线程Thread(row,col)负责计算C(row,col)
	if (row < _wa&&col < _wa)
	{
		double sum = 0;
		for (int i = 0; i < _wa; ++i)

			//_A[row*_wa+i]索引_A的一行，_B[i*_wb+col]索引_B的一列

		{
			sum += _A[row*_wa + i] * _B[i*_wb + col];

			
		}
		_C[row*_wa + col] = sum;
	}


}
void MatrixMultiplication_CUDA1(const double* M, const double* N, double* P, int _wa, int _ha, int _wb)
{
	cudaSetDevice(0);  //设置目标GPU   

	double *Md, *Nd, *Pd;
	
		int size_a = _wa * _ha * sizeof(double);//字节长度  
	int size_b = _wb * _wa * sizeof(double);//字 节长度
	int size_c = _ha * _wb * sizeof(double);//字节长度

	cudaMalloc((void**)&Md, size_a);
	cudaMalloc((void**)&Nd, size_b);
	cudaMalloc((void**)&Pd, size_c);

	//Copies a matrix from the memory* area pointed to by src to the memory area pointed to by dst  
	cudaMemcpy(Md, M, size_a, cudaMemcpyHostToDevice);
	cudaMemcpy(Nd, N, size_b, cudaMemcpyHostToDevice);

	
	dim3 dimGrid(WIDTH / TILE_WIDTH, HEIGHT / WIDTH);
	dim3 dimBlock(TILE_WIDTH, TILE_WIDTH);
	
	MatrixMulKernel << < dimGrid, dimBlock >> >(Md, Nd, Pd, _wa, _wb);

	cudaMemcpy(P, Pd, size_c, cudaMemcpyDeviceToHost);
	//释放设备上的矩阵  
	cudaFree(Md);
	cudaFree(Nd);
	cudaFree(Pd);
}
int main()
{
	double A[16] = { 1,2,3,4 ,1,2,3,4,1,2,3,4,1,2,3,4};
	double B[16] = { 5,6,7,8,5,6,7,8,5,6,7,8,5,6,7,8};
	double C[] = {0};

	
	printf("A矩阵\n");
	for (int i = 0; i < 16; i++)
	{
		if (i % 4== 0)
		{
			printf("\n");
		}
		printf("  %5f", A[i]);
	}
	printf("\n");
	printf("B矩阵\n");
	for (int i = 0; i < 16; i++)
	{
		if (i % 4 == 0)
		{
			printf("\n");
		}
		printf("  %5f", B[i]);
	}
	printf("\n");
	printf("结果C矩阵\n");
	MatrixMultiplication_CUDA1(A, B, C, 4, 4, 4);

	for (int i = 0; i < 16; i++)
	{//输出矩阵样式
		if (i % 4 == 0)
		{
			printf("\n");
		}
		
		printf("  %f", C[i]);
	}
	
	system("pause");
	return 0;
}