Hi Ilteber,

We did some scaling tests of the compressible flow solver in Nektar++ and found that the code gave good scaling results down to 1000 degrees of freedom (DOF) per core. For a lower number of DOF per core the code started losing scalability. In our case we had 48 cores per node and around 1.8 GB per core.

You can see the number of DOF of your problem using:

$ FieldConvert -m dof <mesh_file>.xml <session_file>.xml out.stdout

Cheers,

Margarida

On Tue, 25 Oct 2022 at 12:34, İlteber Özdemir <rilteber@gmail.com> wrote:

This email from rilteber@gmail.com originates from outside Imperial. Do not click on links and attachments unless you recognise the sender. If you trust the sender, add them to your safe senders list to disable email stamping for this address.

 

Hello all,

Is there a study on the optimum amount of nodes per core in Nektar++? What is the recommended number of elements per core when using the solvers in parallel, and how is this affected by the order of the solution used? The same goes for memory used too: what is the recommended memory when using Nektar++ in parallel?

Thank you in advance for the help!  

Best regards,

Ilteber

--

İlteber R. Özdemir

_______________________________________________
Nektar-users mailing list
Nektar-users@imperial.ac.uk
https://mailman.ic.ac.uk/mailman/listinfo/nektar-users

--

Margarida Moragues Ginard

MSCA-IF Researcher
BCAM - Basque Center for Applied Mathematics
Alameda de Mazarredo, 14
E-48009 Bilbao, Basque Country - Spain
Tel. +34 946 567 842

mmoragues@bcamath.org | www.bcamath.org/mmoragues

(matematika mugaz bestalde)