Nov 25, 2017

Using openARC for CUDA GPUs

matmul OpenACC-CUDA example:

export OMP_NUM_THREADS=2
./matmul_CUDA 
Number of OpenMP threads 2
CPU Elapsed time = 0.181340 sec
Accelerator Elapsed time (CUDA) = 0.006136 sec
Accelerator Elapsed time (OpenACC) = 0.006496 sec
Verification Successful err = 0.000000e+00


Followers