Hi everyone,
I installed PyFR successfully for bench-marking purpose. First I installed it with all its dependencies including CUDA and OPENCL.
When I run it , it was started executing on CPU + GPU , and it ran successfully.
But now I want to run it on CPU only, so I installed it on a multi-cores machine which is not having any GPU card, without OPENCL and CUDA dependencies.
When I am trying to run it, with " mpirun -n 4 pyfr run couette_flow_2d.pyfrm couette_flow_2d.ini" command
it gives following error :
ImportError: No module named ‘pycuda’
So I want to know , how to run it on multi-cores with MPI.
Regards,
Roshan
Hi Roshan,
As per the documentation to run PyFR on a CPU using the OpenMP backend
you need to pass -b openmp to $ pyfr. Otherwise, PyFR will run using
the CUDA backend on an NVIDIA GPU.
Regards, Freddie.
Thank you Freddie,
It worked, thanks for your suggestion.
I am trying to benchmark PyFR , please suggest some datatsets or provide me the link for that, so I can take benchmark of PyFR.
waiting for your reply.
Regards,
Roshan
Hi Roshan,
Suitable sample data sets and configuration files can be found in the
following series of posts on the mailing list:
example datasets
(Look for the post by Jin Seok.)
Thank you Freddie !!! once again …
Regards,
Roshan