[Mitgcm-support] Re: success

[mitgcm-support at dev.mitgcm.org]

Cool. There are some starting parameters on the atl6 CD's.
Also at Arne's web site. You will beed biharmonic and timestep
of around ten minutes would probably be a good start.
CFL condtions for biharmonic is very approx.

   k4.dt/L^4 < 0.1

 where k4 is biharmonic coeff,
       dt is timestep,
       L  is grid spacing.
 CFL is worse for the smallest L on your grid.
 We can scale k4 a bit in the small L regions later
but a quick test first would be useful and good for morale!

Suggest first run is one with just hydrography, no forcing,
no relaxation. Run for one month and see what happens!
You can ask the BU people for more space under /project/

Be in later.

Chris
Stephanie Dutkiewicz wrote:
> 
> Hi Chris -
> Hope Helen is doing okay! Give her my love.
> 
> The model ran for 50 timesteps (no forcing, no stratification) on
> both the 32 and 64 processes.
> 
> 32-processors:
> took 6 minutes
> used ~7500 sec CPU time
> and 3.3G of memory
> 
> 64-processors:
> took 4 minutes
> used ~7500 sec CPU time
> and 4.7G of memory
> 
> However, I overfilled the disc space quota with only 1 dump, and 1 set of
> up pickups: but that was for each run, so maybe this will be okay.
> 
> Need now to talk about viscosity (biharmonic? what sort of magnitude)
> and timestep (I had run the above runs with 20minutes - but with nothing
> happening that doesn't really help us at all).
> 
> ps. nothing did happen.... just lots of zeros, and there are several
> tiles with no water on them...
> 
> pps. I have the hydrographies and wind ready, so could potentially run
> a few time steps tooinght?