OpenMPを用いてマルチGPU

  1
  2
  3

int GPU_N;
cudaGetDeviceCount(&GPU_N);
printf("GPU Device Count:%d\n",GPU_N);

  1
  2
  3
  4
  5
  6
  7
  8
  9
 10
 11
 12
 13
 14
 15
 16
 17
 18
 19
 20
 21
 22
 23
 24
 25
 26
 27

float *d_f0,*d_f1;
 
#pragma omp parallel num_threads(2)
#pragma omp sections
{
    #pragma omp section //GPU0
    {
      cudaSetDevice(0);
      cudaMalloc( (void**)&d_f0, sizeof(float)*512);
            #pragma omp barrier
            #pragma omp barrier
          }
    #pragma omp section //GPU1
    {
      cudaSetDevice(1);
      cudaMalloc( (void**)&d_f1, sizeof(float)*512);
            #pragma omp barrier
            #pragma omp barrier
          }
}