Runtime Break: Access violation reading location cudaMalloc runtime error

Accelerated Computing CUDA CUDA Programming and Performance
1

Dear Everyone,

I have a runtime problem,

I post 2D small arrays to kernel, there is no problem in building this project (There was a lot of problems, that has been solved.),
But in runtime, coressponding to each cudaMalloc one exception occurs and Break Message appears.

my code is:

[font="Courier New"]


    int random [iter_end_1][iter_end_2]; // 3*15 array
    int res [iter_end_1][iter_end_2]; // 3*15 array

    int * result ;
int res_size = iter_end_1 * iter_end_2 * sizeof (int);

int * rand;
int rand_size =  iter_end_1 * iter_end_2 * sizeof (int);

cudaMalloc( (void**)&rand, rand_size );                                                    // exception occurs but not break message
cudaMemcpy( rand, random, rand_size, cudaMemcpyHostToDevice );          // both exception and break message occur

cudaMalloc( (void**)&result, res_size );
cudaMemcpy( result, res, res_size, cudaMemcpyHostToDevice );  

    // kernel call

    [/font]