Hi Zhenyang
I wonder if you can share the comparison between pyFR and Nektar++ (and Nek5000 maybe).
I am very interested as I am evaluating which code I use as a high-order CFD tool.
Kind regards
syavash
Hi Syavash,
That was long time ago. And I mainly uses PyFR compressible solver now. But we do have some plans to compare PyFR incompressible solver with Nek5000 or Neko in the future.
Zhenyang
I see.
Have you published any work with pyFR?
I have performed simulations using Nektar and OpenFOAM. But I find pyFR very alluring in terms of accuracy and speedup.
Could you provide me some insight as someone experienced with Nektar and pyFR, about their overall performances?
Great thanks
syavash
We have some large scale simulation but not publish yet. I mainly uses GPU backend to accelerate my simulations so it is hard to have a direct comparison with Nektar. We really appreciate GPU acceleration feature of PyFR. We chose PyFR at the beginning based on this point. But anyway, put those high order solvers together to have a comparison would be very interesting.
Zhenyang
1 Like
Hello, I see that you are interested in simulating pyfr on a larger scale. I would like to know how large this scale is and how many cards are planned to be used? Are there any plans to use platforms other than NVIDIA?
Currently the largest one is around 500million grids with around 160 AMD cards
wow , Is it a supercomputer in Europe? What is the problem calculated?
yeah it is on LUMI. I am interested in aeroacoustic problems so domain is always fairly big. The big case is related to jets.
Can all of this 'aeroacoustic problem ’ be simulated in pyfr?