#include <cstdio>

const int Blocks = 5;

__global__ void kernel( int * a ) 
{
    int i = blockIdx.x;
    a[i] = i;
}

int main( void ) 
{
    int a[Blocks], *d_a;
    int size = sizeof(int) * Blocks;

    // Allocate d_a on the device
    cudaMalloc((void**) &d_a, size);

    // Fill d_a with the id numbers
    kernel<<<Blocks,1>>>(d_a);

    // Copy d_a onto a on the host
    cudaMemcpy(a, d_a, size, cudaMemcpyDeviceToHost);

    // Output the results
    for (int i = 0; i < Blocks; i++)
        printf( "Hello world from %d!\n", a[i]);

    return 0;
}