kimura/main.cu at main · githubsands/kimura · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
#define N 512
int main(void) {
  int *a, *b, *c;
  int *d_a, *d_b, *d_c;
  int size = N * sizeof(int);
  cudaMalloc((void **)&d_a, size);
  cudaMalloc((void **)&d_b, size));
  cudaMalloc((void **)&d_c, size));
  a = (int *)malloc(size); random_ints(a, N);
  b = (int *)malloc(size); random_ints(b, N);
  c = (int *)malloc(size);
  cudaMemcpy(d_a, a, size, cudaMemcpyHostToDevice);
  cudaMemcpy(d_b, b, size, cudaMemcpyHostToDevice);
  add<<N,1>>>(d_a, d_b, d_c);
  add<<N,1>>>(d_a, d_b, d_c);
  cudaMemcpy(c, d_c, size, cudaMemcpyDeviceToHost);
  free(a); free(b); free(c);
  cudaFree(d_a); cudaFree(d_b); cudaFree(d_c);
  return 0;
}

__global__ void add(int *a, int *b, int *c) {
  c[threadIdx.x] = a[threadIdx.x] + b[threadIdx.x];
}