cuda向设备端传递多个一维数组，并在核函数修改数组元素值

Question

现在有多个一维数组，我想将他们拷贝到设备端修改他们的值，然后拷贝回host端，代码如下：
#include "cuda_runtime.h"
#include "device_launch_parameters.h"
#include "helper_cuda.h"
#include
#include
using namespace std;
#define N 10
global void gpuKernel(int ptr)
{

int tid=threadIdx.x;
int bid=blockIdx.x;
if (tid<N &&bid<N)
{

ptr[bid][tid]=N;

}
}
int main()
{
int **hptr,hostArray;
int dptr,devArray;
hptr=new int [N];
cudaMalloc(&dptr,N*sizeof(int));

hostArray=new int *[N];
devArray=new int *[N];
for (int i=0;i<N;i++)
{
hostArray[i]=new int[N];
for (int j=0;j<N;j++)
{
hostArray[i][j]=1;
}
}
for (int i=0;i<N;i++)
{
cudaMemcpy(devArray[i],hostArray[i],N*sizeof(int),cudaMemcpyHostToDevice);
hptr[i]=devArray[i];//保存每个数组的头指针
}

cudaMemcpy(dptr,hptr,N*sizeof(int*),cudaMemcpyHostToDevice);    
dim3 dimblock=N;
dim3 dimgrid =N;
gpuKernel<<<dimgrid,dimblock>>>(dptr);
for (int i=0;i<N;i++)
{       cudaMemcpy(hostArray[i],devArray[i],N*sizeof(int),cudaMemcpyDeviceToHost);      
}
for (int i=0;i<N;i++)
{
    for (int j=0;j<N;j++)
    {
        cout<<hostArray[i][j]<<",";
    }
    cout<<endl;

}
//free memory
for (int i=0;i<N;i++)
{
    delete []hostArray[i];
    cudaFree(devArray[i]);
}
cudaFree(dptr);
/*cudaFree(devArray);*/
delete[]hptr;
delete []hostArray;
delete[]devArray;   

return 0;

}
最终结果是元素值都为修改，纠结这个问题几天了，并且仔细拜读bendanban大神的博客http://blog.csdn.net/bendanban/article/details/7669624，可能自己理解有误，望各位大神指点