[MITgcm-devel] two of my favorite packages: exf/cal

[Dimitris Menemenlis]

Martin, are you trying multi-category ice?  Are you starting from rest?
The first 3 days are always unstable when you start from rest.  So maybe
reduce time step size even further?  20 s?

The other option is to try starting from one of the standard pickup files with 
multicategory turned off, to see if the code, etc., works OK.  I recommend

ftp://ecco2.jpl.nasa.gov/data1/cube/run_template/pickup.0000000216.cube49
ftp://ecco2.jpl.nasa.gov/data1/cube/run_template/pickup_seaice.0000000216.cube48

A third option is to try one of the 32x6x32 setups first, the one that is 
preceded by

# 32*32*6*15, 1-cpu cube sphere

or by

# 32*32*6*15, 2-cpu cube sphere on the altices with ifort

in http://mitgcm.org/cgi-bin/viewcvs.cgi/MITgcm_contrib/high_res_cube/README_ice

to make sure that everything works OK first before working on.

Let me know how things go.  If needed I can also experiment with reproducing 
your crash on the altix.

D.