C Structures

Accelerated Computing CUDA CUDA Programming and Performance
1

I am trying to use structs in my CUDA program, however I seem to be missing something important here. I have simplified the following code to exploit the problem:

struct genericArray

{

	float* values;

};

void runTest()

{

    genericArray h_idata;

   // allocate host memory

    h_idata.values = (float*) malloc( sizeof( float) * N);

   // initalize the memory

    for( unsigned int i = 0; i < N; ++i) 

    {

        h_idata.values[i] = (float) i;

    }

   // allocate device memory

    genericArray d_idata;

    CUDA_SAFE_CALL( cudaMalloc( (void**) &(d_idata.values), sizeof( float) * N));

   // copy host memory to device

    CUDA_SAFE_CALL( cudaMemcpy( &(d_idata.values), &(h_idata.values), sizeof( float) * N, cudaMemcpyHostToDevice) );

}

I have omitted the rest of the method runTest for brevity.

The error occurs at the final call to cudaMemcpy. My program crashes with an error:

Can anyone tell me what’s going wrong here?

2

You are passing memcpy the address of the pointers, you just need to pass the pointers.

You have to pass malloc the address of the pointer so that it can alter the value of the pointer itself.