search

CICE-Consortium / CICE

Development repository for the CICE sea-ice model
Other
57 stars 131 forks source link

C grid computational efficiency #704

Open JFLemieux73 opened 2 years ago

JFLemieux73 commented 2 years ago

We can list ideas of what we think could be done to reduce cpu time.

-in div_stress there are many calculations such as dxT(i,j)**2...can we store these as new arrays? -if visc_method='avg_zeta', the halo update of the ice strength is probably not needed. Same idea if visc_method='avg_strength' I think the halo update of the viscosities is not needed... -there are unnecessary calculations in strain_rates_U if visc_method='avg_zeta'.

JFLemieux73 commented 2 years ago

One thing also to consider is that if you sum up the number of u an v that are calculated, this number is ~5% larger for the C grid (estimation from GLOBAL gx3 simulations). This is more pronounced (~10%) for the Northern hemisphere (more channels...e.g. in the CAA).

FYI @apcraig @eclare108213 @dabail10