[MITgcm-devel] USE_FACTORIZED_POLY in find_rho

[Jean-Michel Campin]

Hi Martin,

In principle, the same option (USE_FACTORIZED_POLY or a different name) 
could be used for all EOS, but since USE_FACTORIZED_POLY is defined 
at the top of the 2 files (find_rho.F & find_alpha.F) 
the unique CPP option would have to be defined by default, which is not 
what we want. So will have to go with 2 different CPP options.
And regarding the name, would prefer something with EOS in it rather
than DENSITY_FORMULAE.

Cheers,
Jean-Michel

On Fri, Jan 20, 2012 at 04:47:57PM +0100, Martin Losch wrote:
> Hi there,
> 
> I use factorized formulae for density for the other EOS types, because they improve the performance on the SX8 (but potentially will also do that on other platforms). So far I have them (in my copy) within #ifdef TARGET_NEC_SX, but maybe I should check them in with a different flag name (e.g., USE_FACTORIZED_DENSITY_FORMULAE)? What do you think?
> 
> There is a USE_FACTORIZED_POLY defined at that top of find_rho.F
> 
> Martin
> 
> PS. These factorized formulae do change the results, so they probably shouldn't be default.
> 
> 
> _______________________________________________
> MITgcm-devel mailing list
> MITgcm-devel at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-devel