Compute number of CUDA blocks given grid size and number of threads per block.

compute-cuda-blocks.cpp

int compute_blocks(int length, int threads_per_block) {
    //integer division:
    //if length is evenly divisable by the number of threads
    //is equivalent to length / threads_per_block, if not
    //it is equivalent to length / threads_per_block  + 1
    return (length + threads_per_block - 1) / threads_per_block;
}

dim3 compute_blocks(int xsize, int ysize, int zsize,
                    int threads_per_block_x,
                    int threads_per_block_y,
                    int threads_per_block_z) {
    return dim3(compute_blocks(xsize, threads_per_block_x),
                compute_blocks(ysize, threads_per_block_y),
                compute_blocks(zsize, threads_per_block_z));
}