Always blowing-up at the same timestep

[wzhu]

Hi everyone,
I am running a Roms-sediment model, but it is always blowing-up at the same timestep with a only message: Blowing-up: Saving latest model state into RESTART file, although I reduce the number of threats and timestep.
The same model was run successfully on another machine with mpirun, I use openmp on a shared-memory machine, could it be the reason?
How can I figure out the reason of blowing-up?

Please help me, any help will be highly appreciated!

Wenwu

[kate]

I would start by looking at the restart file with ncview. If it says it wrote to record 3 while blowing up, focus on record 3. Find out which field went bad and where it went bad. Can you try a serial run? Can you try a different compiler? How about optimized vs. unoptimized? Which compiler are you using now?

[wzhu]

I would start by looking at the restart file with ncview. If it says it wrote to record 3 while blowing up, focus on record 3. Find out which field went bad and where it went bad. Can you try a serial run? Can you try a different compiler? How about optimized vs. unoptimized? Which compiler are you using now?

Thanks for your timely reply, Kate.
I will try that and feed back.