Full restart

[FieldingChase]

Is there an NekRS analogue of Nek5000's full_restart subroutine?

[stgeke]

No but you could do this through the nek5000 interface. What are you after which requires an accurate restart?

[FieldingChase]

I am doing time-averaging of a turbulent channel flow for turbulence statistics. When I was running this simulation in Nek5000 on a cluster, the simulation would hit the wall-time and stop before it had run for as long as I intended. I did the full restart to affect the averaging over time as little as possible when I averaged both runs together. I might not need this full restart when I run NekRS on GPUs. Anyway, I will take a look at doing this full restart through the Nek5000 interface.

Thanks for your help.

[stgeke]

I don't think for this application you would need an accurate restart.

[stgeke]

btw: if you specify in par

checkpointInterval = 0 # dump field file at the end only
stopAt = elapsedTime
elapsedTime = 55 # 5 min smaller than job time to account for time spent during setup

you can ensure you'll get a checkpoint file at the end of your run.

[FieldingChase]

Okay, thanks for your assistance.