[MITgcm-support] Re: MIT model (c19) from JPL

heimbach at mit.edu heimbach at mit.edu
Wed Feb 18 09:18:04 EST 2004


Hi Elena,

the JPL code based on c19 is fairly old and has 
considerably departed from the MITgcm main version
so it's a bit hard for me to judge.

Nevertheless, one immediate guess is the 
big/little Endian issue for PC clusters.
Here are a few things you could try:
* if you're compiling with g77 or ifc, add the following CPP option
  to your DEFINES list: -D_BYTESWAPIO
* if you're compiling with pgf77, add the following compiler option
  to your FFLAGS list: -byteswapio
A collection of CPP & compiler options for various platforms
can be found at 
http://dev.mitgcm.org/cgi-bin/viewcvs.cgi/MITgcm/tools/build_options/

If you find out that that's not the problem,
let us know and give more details as to what goes wrong.

Cheers
-Patrick


Quoting eyulaeva at ucsd.edu:

> Patrick,  hi:
> I was corresponding with you a year ago on the MIT model. Now I have
> another problem
> I am trying to run the pre-release version of the MIT model (c19) from JPL
> on the Linux PC cluster at Scripps. The model runs fine on cluster
> machines, however the output fields  are different from  the ones obtained
> at  IRIX64 machine at JPL  and are not realistic.
> I have checked everything I could have thought of , but didn't find a
> solution.  Do you know what can be getting wrong?
> 
> 
> Thank you very much
> I would really appreciate your help
> 
> Elena
> 
> 






More information about the MITgcm-support mailing list