Thanks a lot,
I didn't use the configure_mpich2 script but managed to make a
few tweaks in mpif90 script provided with mpich2 installation by package
manager. The original script was calling for intel compilers by default.
Modifying it certainly solved the problem.
Asmi
On Wed, Apr 25, 2012 at 5:52 PM, David A Case <case.biomaps.rutgers.edu>wrote:
> On Wed, Apr 25, 2012, Asmita Gupta wrote:
>
> > I have run all the bugfixes as well as ./AT15_Amber11.py by hand. Also
> > installation of serial version was successfull.
> >
> > The configuration command was:
> > ./configure -mpi gnu
> >
> > which mpif90 gives this:
> > /opt/mpich2-1.4.1p1/gcc/bin/mpif90. (so is it still a compiler problem??)
>
> Yes: it is almost certainly a compiler problem (see below).
> >
> > mpif90 -c -O3 -ffree-form -o constants.o _constants.f
> > ifort: command line warning #10006: ignoring unknown option '-ffree-form'
> > _constants.f(89): error #5149: Illegal character in statement label field
> > [m]
> > module constants
> > ^
>
> Your mpif90 program is calling ifort (the Intel compiler), but you
> configured
> amber with the gnu compilers. Your MPI installation must use the same
> compiler as you used for configure. So you can configure with intel, or
> change your MPI. One recommended way to do the latter is to use the
> configure_mpich2 script in $AMBERHOME/AmberTools/src.
>
> ...dac
>
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Apr 26 2012 - 05:30:03 PDT
