I concur with Ross, just recompile MPI (whatever flavor you prefer)
with the compilers you intend to use with Amber, put it in your
home directory and then link only against those libraries.
I've gotten so that I automatically do that on a new system.
Simplifies life a lot.
best,
jim
On Tue, 2005-02-22 at 09:54, Ross Walker wrote:
> Dear Frank,
>
> > I'm trying to compile the parallel version of sander8 on a Linux
> > cluster. The MPI library on the node seems to have second trailing
> > underscore:
> >
> > nm /usr/local/mpi/lib/libmpich.a | grep mpi_init
> > 00000000 W mpi_init__
> > 00000000 T pmpi_init__
> > 00000000 W mpi_initialized__
> > 00000000 T pmpi_initialized__
> >
> > However the package compiled with either ifort or ifc (the
> > Intel Fortran
> > compilers version 8 or 7) only has one trailing underscore, which
> > results in linking error. I was wondering if there's any way to
> > reconcile this discrepancy without recompiling the MPI.
>
> This is a common problem when the mpi libraries were compiled with a
> different compiler to the program you want to run. Typically on Linux
> machines people compile the mpi libraries with GCC and g77. Hence you cannot
> link against ifort F90. Aome compilers support flags such as
> -Mnosecondunderscore but my advice would be to download the latest version
> of either mpich or lam and compile it yourself (statically if you can) using
> the same compiler as you are using to compiler sander.
> Make sure you enable fortran 90 support when compiling the mpi libraries.
> Adjust your paths so that the version you have compiled in your home
> directory is used in place of the version installed in /usr/local. This way
> you will have much more control on how things are compiled and fewer
> hassles.
>
> All the best
> Ross
>
> /\
> \/
> |\oss Walker
>
> | Department of Molecular Biology TPC15 |
> | The Scripps Research Institute |
> | Tel:- +1 858 784 8889 | EMail:- ross.rosswalker.co.uk |
> | http://www.rosswalker.co.uk/ | PGP Key available on request |
>
> Note: Electronic Mail is not secure, has no guarantee of delivery, may not
> be read every day, and should not be used for urgent or sensitive issues.
>
>
> -----------------------------------------------------------------------
> The AMBER Mail Reflector
> To post, send mail to amber.scripps.edu
> To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
--
James W. Caldwell 650-724-5322
Department of Chemistry
Clark Center, S2.2, Rm S294
Stanford University
Stanford, CA 94305-5447
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
Received on Tue Feb 22 2005 - 20:53:00 PST
