[petsc-dev] PETSC's compilation

[David sheehan]

I try openmpi, there are the following error,

./configure --prefix=/home/David/petsc300p11_openmpi --with-cc=gcc
--with-cxx=g++ --with-fc=ifort --download-openmpi=1 --download-hypre=1
=================================================================================
             Configuring PETSc to compile on your system
=================================================================================
TESTING: check from
config.libraries(config/BuildSystem/config/libraries.py:126)
*********************************************************************************
         UNABLE to CONFIGURE with GIVEN OPTIONS    (see configure.log for
details):
---------------------------------------------------------------------------------------
Currently --download-openmpi option does not work with --prefix install.
Suggest installing OpenMPI separately, and then configuring PETSc with
--with-mpi-dir option.
*********************************************************************************

How to fix it? thanks.

David



On Mon, Mar 22, 2010 at 4:00 PM, Satish Balay <balay at mcs.anl.gov> wrote:

> If you need openmpi - try:
>
> ./configure --with-cc=gcc --with-cxx=g++ --with-fc=ifort
> --download-openmpi=1 --download-hypre=1
>
> your current install of openmpi [which is probably installed with g77]
> is not useable from ifort.
>
> Satish
>
>
> On Mon, 22 Mar 2010, David sheehan wrote:
>
>  > You mean, I can build up PETSC with openmpi plus ifort? if so, how to
> set up
> >
> > the "configure" in details, thanks.
> >
> > David
> >
> > On Mon, Mar 22, 2010 at 3:26 PM, Matthew Knepley <knepley at gmail.com>
> wrote:
> >
> > > Yes, the fortran name mangling for the MPI Fortran bindings is wrong.
> You
> > > must
> > > always build everything with the same compilers, which is why packages
> do
> > > not
> > > mix with code you build yourself. Looks like you will have to use
> > > --download-openmpi.
> > >
> > >    Matt
> > >
> > > On Mon, Mar 22, 2010 at 3:23 PM, David sheehan <
> david.sheehanjr at gmail.com>wrote:
> > >
> > >> Hi,
> > >> I am trying to compile the PETSC with hypre. I have Intel FORTRAN,
> openmpi
> > >> and  lam
> > >> MPI  (gcc version 3.4.6 (Red Hat 3.4.6-3)) available.
> > >>
> > >> Since my application code is in FORTRAN with dynamic memory allocation
> in
> > >> the code,
> > >> so I have to use Intel FORTRAN(ifort) as the compiler to link PETSC
> with
> > >> my application
> > >> code. I can build up PETSC(petsc-3.0.0-p11) plus hypre with Lam MPI
> but
> > >> without ifort
> > >> successfully. However I can not use ifort to link the PETSC with my
> > >> application success-
> > >> fully. The link always shows erros about "undefined reference" to
> MPIparts in my application
>  > >> code such as 'mpi_send_', 'mpi_recv_', 'mpi_waitany_',
> 'mpi_get_count_',
> > >> and etc.
> > >>
> > >> Can any one help me out? thanks in advance.
> > >>
> > >> Best Regards,
> > >>
> > >> David
> > >>
> > >
> > >
> > >
> > > --
> > > What most experimenters take for granted before they begin their
> > > experiments is infinitely more interesting than any results to which
> their
> > > experiments lead.
> > > -- Norbert Wiener
> > >
> >
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20100323/e6b1f42c/attachment.html>