[cam-users] pgf90 compiler differences
David Noone
dcn@gps.caltech.edu
Tue, 15 Jul 2003 17:05:24 -0700 (PDT)
Hi Norm,
I played with a few different linux compilers (pgf 4.0, intel 5, 6 and
7, absoft) with CCM3, and as you say, there are some difference. I
believe one should expect differences to the limit of numerical
precision (as the intrinsic math libraries, etc., will do the math ever
so slightly differently) - recall there is some discussion of this in
the users guide. Does this sound like what you see? It might not be
surprising that you find the biggest differences in the output from
schemes that have somewhat stronger dependence on the numerical round
off than the core dynamics. The precip from a _single_ (and the first)
time step will be particularly sensitive. One test might be to use the
"adiabatic physics" (?).
In general, these small errors with amplify until the "weather" diverges
(exactly like the attractor Ed Lorenz found), and a comparison of fields
be comes tricks. The real test is if the statistical properties of the
climate is the same with different compilers. Still, it's worth trying
to make sure that this is what's happening with the upgrade to pgf 5,
or if it is something more substantial.
David
On Tue, 15 Jul 2003, Norm Wood wrote:
> Hi,
>
> In the last few days we've upgraded (?) our pgf90 compiler
> to version 5.0-1 from 4.0-1 and I've done a pair of test
> runs with CAM 2.0.2. Maybe not surprisingly, there are
> differences in the NSTEP=0 results for the model compiled
> with the different versions of the compiler (from running
> cprnc on the h0 output). The differences occur mainly in
> variables related to moist processes (e.g. CME, PRECC, PRECL,
> LWCF, SWCF) but VTGWORO and UTGWORO differ also.
>
> Before I dig further, I thought I'd ask whether any one
> else has tested the new version and looked into the differences.
>
> I realize that 5.0 isn't officially supported - this should
> be a heads-up to anyone else looking at upgrading their pgf90
> compiler.
>
>
> Norm
>
> --
> Norm Wood
>
> Department of Atmospheric Science
> Colorado State University
> norm@atmos.colostate.edu
> phone: 970-491-8587
>
> _______________________________________________
> cam-users mailing list
> cam-users@ucar.edu
> http://mailman.ucar.edu/mailman/listinfo/cam-users
>
>
--
--------------------------------------------------------------------
California Institute of Technology, Pasadena, CA, 91125, USA
Division of Geological and Planetary Sciences
Mail Stop 100-23, Phone (626) 395-6982
http://www.gps.caltech.edu/~dcn