CUDA sample

cudatest.cu

#include <stdio.h>
#include <cuda.h>

__global__ void
func(void){
  printf("%d %d %d\n",blockIdx.x, blockDim.x, threadIdx.x);
}

int
main(void){
  cudaSetDeviceFlags(cudaDeviceMapHost);
  int const n_thread = 3;
  int const n_block = 4;
  func<<<n_block,n_thread>>>();
  printf("Version = %d\n",CUDA_VERSION);
  //cudaThreadSynchronize();
}