Hi Jason,
You are right. mpicc, mpiCC and mpif90 is indeed wrappered by athena and
kraken, which will probaly make the pnetcdf failed during the initial
configuration. But you can download mpich-2.1.1.tar.gz from mpich website,
then compiled with any version of pgi you will get mpicc, mpiCC, mpif90 and
mpirun in your home directory. As long as the enviroment is correctly setup,
it works fine to use ./configure -mpi pgi; make parallel; make
test.parallel. However, never use this mpirun for other usage except test
since the athena and kraken may not allow us due to the policy, That's why
mpi*** is alway hidden by the athena and kraken.
Best Regards,
Lin
2011/3/17 Jason Swails <jason.swails.gmail.com>
> The problem with trying it on Athena or Kraken is that they have their own
> software environment, and invoke compilers through wrappers (though I
> suppose I could always try just using pgi straight-out, but then MPI won't
> work since there's no mpif90 or mpicc). I'll see what I can do.
>
> Thanks,
> Jason
>
> On Thu, Mar 17, 2011 at 10:40 AM, Lin Fu <linfu3200.gmail.com> wrote:
>
> > Hi Ben,
> >
> > Thank you for your answer. We tried it on the athena and kraken, which
> have
> > all kinds version of PGI compilers. Ben, do you have an account on the
> > athena or krakren. Dwight once told me you guys also use teragrid
> resource,
> > right? I will also try to figure it out if time allowed.
> >
> > Best Regards,
> > Lin
> >
> > 2011/3/17 Ben Roberts <ben.roberts.geek.nz>
> >
> > > Hi Lin,
> > >
> > > On 17/3/2011, at 2:31 a.m., Lin Fu wrote:
> > >
> > > > As for the mtkpp, there are lots of compiler error with pgi even if
> the
> > > removing of "-Wall" and some modification of head file, which invovles
> > > redefintion of some variables, which has been already defined in the
> > > include/ folder of pgi installation directory. Hope Ben will give patch
> > > about it.
> > >
> > > Neither I nor Martin Peters (the chief MTK++ developer) have access to
> > the
> > > PGI compilers. Nor do we know what, precisely, the errors are that
> you've
> > > been receiving. At present, then, we can't easily help you.
> > >
> > > Do you have administrative control over a machine with the PGI
> compilers
> > > installed on it? If so, and if you would like us to help you get MTK++
> up
> > > and running with the PGI compilers, perhaps you could provide us with a
> > > user-level account on the machine in question. If you're able and
> willing
> > to
> > > do that, please e-mail me off-list.
> > >
> > > --
> > > For greater security, I support S/MIME encryption.
> > >
> > >
> > > _______________________________________________
> > > AMBER-Developers mailing list
> > > AMBER-Developers.ambermd.org
> > > http://lists.ambermd.org/mailman/listinfo/amber-developers
> > >
> > >
> > _______________________________________________
> > AMBER-Developers mailing list
> > AMBER-Developers.ambermd.org
> > http://lists.ambermd.org/mailman/listinfo/amber-developers
> >
>
>
>
> --
> Jason M. Swails
> Quantum Theory Project,
> University of Florida
> Ph.D. Candidate
> 352-392-4032
> _______________________________________________
> AMBER-Developers mailing list
> AMBER-Developers.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber-developers
>
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Sat Mar 19 2011 - 01:30:03 PDT
