#include<stdio.h>
#include<math.h>
#include<cutil.h>
#include<addone_kernel.cu>

int main(int argc, char ** argv)
{
   int p = 256;  // threads per block
   int lengthvector = 10240;
   dim3 threads(p,1,1);  // threads per block
   dim3 grid(lengthvector/p,1,1);  // lengthvector should be multiple of p
   // numbers of blocks  to cover array
   int sharedMemSize = 0;
   unsigned int memSize = lengthvector*sizeof(float);

   // vector on host
   float* h_vector = (float*) malloc(memSize);

   // put something in the vector 
   for(int i=0;i<lengthvector;i++) {
     h_vector[i] = (float)(i%10);
   }
   for(int i=0;i<20;i++) { // print out part of vector
     printf("%.1f \n", h_vector[i]);
   }

   float *d_vector;  // vector on device

   // allocate memory on device
   CUDA_SAFE_CALL(cudaMalloc((void**)&d_vector, memSize ));

   // transfer memory host to device 
   CUDA_SAFE_CALL(cudaMemcpy(d_vector, h_vector, memSize, 
          cudaMemcpyHostToDevice));

// call kernel
   addone<<< grid, threads, sharedMemSize >>> (d_vector);

   // transfer memory back 
   CUDA_SAFE_CALL(cudaMemcpy(h_vector, d_vector, memSize, 
          cudaMemcpyDeviceToHost));

   // Check if kernel execution generated an error
   CUT_CHECK_ERROR("Kernel execution failed");

   printf("after\n");
   for(int i=0;i<20;i++) { // print out part of vector
     printf("%.1f \n", h_vector[i]);
   }
  // cleanup
   CUDA_SAFE_CALL(cudaFree(d_vector));
   free(h_vector);
// Use CUDA Utility Tool to exit cleanly
   CUT_EXIT(argc, argv); 

}