#include <stdio.h>
#include <stdlib.h>
#include <stdint.h>

/* nvcc cuda_template.cu -o cuda_template */ 


__global__ void inc(float *v, float *r){ 
  unsigned int gtid = blockIdx.x * blockDim.x + threadIdx.x;
  
  r[gtid] = v[gtid] + 1; 
}

#define N 32

int main(int argc, char **argv){
  float v[32]; 
  float r[32];
  float *dv = 0;  
  float *dr = 0;
 

  //generate input data
  for (int i = 0; i < 32; ++i) { 
    v[i] = 1.0; 
  }

  cudaMalloc((void **)&dv, sizeof(float) * N ); 
  cudaMalloc((void **)&dr, sizeof(float) * N ); 

  cudaMemcpy(dv, v, sizeof(float) * N, cudaMemcpyHostToDevice);

  inc<<<1, 32>>>(dv,dr);
  cudaMemcpy(r, dr, sizeof(float) * N , cudaMemcpyDeviceToHost);

  cudaFree(dv);
  cudaFree(dr);
  
  // show results 
  for (int i = 0; i < 32; ++i) { 
    printf("%f ", r[i]);
  }

}