[MITgcm-devel] heff_max...more sea ice issues

[Dimitris Menemenlis]

> Dimitris: Does this also make sense for the EVP solver?

No.

> We may want to include a flag that allows this only if the LSR solver is 
> used. What do you think?

Definitely.  Will you or I take care of that?

> For example one could have deltaTmom=900 and SEAICE_deltaTdyn=3600, which 
> would call LSR solver only once every 3 model time steps.

My six-year-old tells me that 3600/900 is 4, not 3 ;-)

> In regards to your "why is the ice  model slow? It's a 2D model? I don't get
> it!", has anyone profiled the seaice dynamics to see where the time (17% of
> the run in my case) is going...maybe one would find some surprises...or is it
> just that there are exchanges at every iteration?

Yes.  But worse there are also global_max's, so you get hit on latency, same as 
the cg2d solver.

As I said in earlier message, old B-grid code, without ice-to-ocean stress did 
not blow up at 1/8th and 1/16th resolutions.  But Patrick is right, your setup 
is too expensive for experimenting with all these parameters.  I am working on a 
smaller regional Arctic setup (I promised it to Patrick about half a year ago) 
where I can experiment with all these options.  So hopefully I will have more to 
say early in New Year.  Cheers, D.