About Open MP and cudaSetDevice()

Question

Does anyone know if the following usage of cudaSetDevice is correct? I want to repeatedly call resources created on different devices at any time, in any host thread; is there a way to do this in CUDA?

 cudaSetDevice(0);
 /...create cuda streams and do some memory allocation on gpu.../
 cudaSetDevice(1);
 /...create cuda streams and do some memory allocation on gpu.../
 #pragma omp parallel num_threads(2)
 { 
   int omp_threadID=omp_get_thread_num();
    ....
   if (omp_threadID==0)
   {
    cudaSetDevice(0);
    /...calling streams/memory created on device 0.../
   }
   else
   {
    cudaSetDevice(1);
    /...calling streams/memory created on device 1.../
    }; 
  };

Answer 1

Yes, something like that should work. Make sure that all things you created on device 0, you only use in OpenMP thread 0, and likewise for device 1 and thread 1.

You may also want to look at the CUDA OpenMP Sample Code , which demonstrates how to use OpenMP threads to each manage an individual device.