BookGPU/Chapters/chapter5/code/ex1.cu

   1 #include <gpulab/vector.h>
   2
   3 __global__ void add(double* a, double const* b, int N)
   4 {
   5         int i = blockDim.x*blockIdx.x + threadIdx.x;
   6         if(i<N)
   7                 a[i] += b[i];
   8 }
   9
  10 int main(int argc, char *argv[])
  11 {
  12         int N = 1000;
  13
  14         // Basic CUDA example
  15         double *a1, *b1;
  16         cudaMalloc((void**)&a1, N*sizeof(double));
  17         cudaMalloc((void**)&b1, N*sizeof(double));
  18         cudaMemset(a1, 2.0, N);
  19         cudaMemset(b1, 3.0, N);
  20         int blocksize = 128;
  21         add<<<(N+blocksize-1)/blocksize,blocksize>>>(a1, b1, N);
  22
  23         // gpulab example
  24         gpulab::vector<double,gpulab::device_memory> a2(N, 2.0);
  25         gpulab::vector<double,gpulab::device_memory> b2(N, 3.0);
  26         a2.axpy(1.0, b2); // BLAS1: a2 = 1*b2 + a2
  27
  28         return 0;
  29 }