[CIG-MC] pgi compilers

Bianco, Todd todd_bianco at brown.edu
Tue Nov 9 13:30:21 PST 2010


Hello All,

To Walter: The /etc/redhot-release is CentOS 4.5, so Redhat 4.X, I suppose.

I have the code compiled and running using OpenMPI built on PGI compilers,
but I have to add

#define NULL 0

to Parsing.c and Pan_problems_misc_functions.c. Adding the new definition of
NULL in global_defs.h will cause other parts of the code to generate errors.


I have not tested the code thoroughly for unintended behavior, but I
compared a two parallel runs, one using MPI built on GNU and one using MPI
built on PGI compilers. There are only very small differences between the
output of the different runs, which may be do to the randomization of tracer
locations in the version I use. For example comparing the vertical velocity
field between the two runs, there is a difference of <0.001% in some
locations. I will write to you if I notice any strange behavior with the PGI
compiler, or with the different definitions of NULL.

Regards,
Todd






On Mon, Nov 8, 2010 at 5:20 PM, Walter Landry <walter at geodynamics.org>wrote:

> "Bianco, Todd" <todd_bianco at brown.edu> wrote:
> > Thanks for the responses. I did not install PGI, but from the
> license.info
> > file, this is 7.0-7
>
> Hmm.  What version of the operating system are you using?  It is
> listed in the supported versions at
>
>  http://www.pgroup.com/support/install.htm#release_info
>
> In any case, you can try adding
>
>  #define NULL ((void *)0)
>
> or
>
>  #define NULL 0
>
> to global_defs.h and see if that works.
>
> Cheers,
> Walter Landry
> walter at geodynamics.org
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://geodynamics.org/pipermail/cig-mc/attachments/20101109/b11bdd00/attachment.htm 


More information about the CIG-MC mailing list