[MITgcm-devel] [MITgcm-cvs] MITgcm/tools/build_options CVS Commit

[Jean-Michel Campin]

Hi Patrick,
I've just removed this optfile, but I could put it back if you
think it's important.
Also, we now have testreport output from OpenAD with pgi compiler and MPI (openmpi),
using optfile MITgcm/tools/build_options/linux_amd64_pgf77
both with -ieee (testreport default) and with -fast option, run on acesgrid:
 http://mitgcm.org/testing/results/2014_08/tr_acesgrid-oad_20140810_0/
 http://mitgcm.org/testing/results/2014_08/tr_acesgrid-oad_20140810_1/
that are supposed to be run once a week.
The PBS script has been checked-in as:
 MITgcm/tools/example_scripts/ACESgrid/test_aces_pgi_oad
Please let me know if there is any problem there.
Cheers,
Jean-Michel

On Sun, Aug 10, 2014 at 01:05:27PM -0400, Jean-Michel Campin wrote:
> Hi Patrick and Oliver,
> 
> I am tempted to remove this new optfile (linux_amd64_pgf90)
> to just keep one standard PGI one (linux_amd64_pgf77):
> a) After the few changes I made last Friday in linux_amd64_pgf77,
>  I was able to use it to compile the atm_gray experiment from Contrib
>  (this one uses both fix F90 format and free F90 format),
> b) and with the little addition to genmake2 I've just check-in,
>  I was able to check that it works also with OpenAD,
>  at least for experiment hs94.1x64x5.
> In fact, with these changes (genmake2 & linux_amd64_pgf77), it's
> back to the same compiler FLAGS as in linux_amd64_pgf90 and, 
> without -ieee, the same FOPTIM setting.
> 
> To check that this was working with OpenAD, I used Oliver OpenAD install 
> on acesgrid, by loading the module "openad" and it just worked.
> 
> May be it would be better if the PGI optfile was named "linux_amd64_pgi"
> (but I am not going to change it now).
> 
> Cheers,
> Jean-Michel
> 
> On Mon, Aug 04, 2014 at 11:54:34PM -0400, Patrick Heimbach wrote:
> > Update of /u/gcmpack/MITgcm/tools/build_options
> > In directory forge:/tmp/cvs-serv2441
> > 
> > Added Files:
> > 	linux_amd64_pgf90 
> > Log Message:
> > Optfile for pgf90 that can handle OpenAD-generated code
> > (so far only serial)
> > 
> > 
> > _______________________________________________
> > MITgcm-cvs mailing list
> > MITgcm-cvs at mitgcm.org
> > http://mitgcm.org/mailman/listinfo/mitgcm-cvs
> 
> _______________________________________________
> MITgcm-devel mailing list
> MITgcm-devel at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-devel