printing from CUDA kernel

cuda_print_buffer.cu

template<typename T>
__global__
void print_buffer(const T* buf, uint num_elements){
  for(int i=0; i<num_elements; i++){
    printf("%x ", buf[i]); // print hex values
  }
  printf("\n");
}

template<typename T>
void launch_print_buffer(const T* buf, uint num_elements, cudaStream_t stream){
  cudaDeviceSynchronize();
  check_cuda_error(cudaGetLastError());
  print_buffer<<<1, 1, 0, stream>>>(buf, num_elements);
  cudaDeviceSynchronize();
  check_cuda_error(cudaGetLastError());
}