[DFTB-Plus-User] Fwd: errors compiling dftb+.mpi

[Alex A. Schmidt]

Dear Bálint,

Thanks for the reply. I moved to gfortran 4.9.2 and compilation of dftb+.mpi
went on smoothly.

Compiltation of serial dftb+_1.2.2 with -fopenmp also worked but it
required the
adittion of -lgomp for the non-static case. For the static case, -lgomp is
not required
since libgomp.a from gcc would be added automatically.

Alex

2014-12-18 12:42 GMT-02:00 Bálint Aradi <aradi at uni-bremen.de>:

> Dear Alex,
>
> > I notice the presence of a file named "penmp" in the
> > /usr/local/dftb+/dftb+.mpi-r4473/extlib/scalapackfx/src
> > directory after the occurrence of this error. If I remove the "-openmp"
> > option from the FXXOP list, this error
> > goes away. So, I guess this must have something to do with "make" not
> > passing options correctly around...
>
> The openmp directive is simply wrong. First of all, this is now called
> -fopenmp for gfortran. And second, it should not be present at all in
> the parallel code. Thanks for spotting it!
>
> > second (and more serious) error:
> > =======================
> >
> >
> > nd(msg, size(msg) * len(msg), MPI_CHARACTER, dest, tag0, mycomm%id,
> error0)
> >
>   1
> > Error: There is no specific subroutine for the generic 'mpi_send' at (1)
>
> This is unfortunately a compiler bug / missing F2003 feautre. GFortran
> 4.9.1 should be to compile mpifx, everything prior that does not work.
> Make sure, your mpif90 wraps a recent gfortran compiler.
>
>   Cheers,
>
>    Bálint
>
>
> Ps. I've added you now manually to the mailing list, not sure, why your
> subscription did not work.
>
> --
> Dr. Bálint Aradi
> Bremen Center for Computational Materials Science, University of Bremen
> http://www.bccms.uni-bremen.de/cms/people/b_aradi/
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.zfn.uni-bremen.de/pipermail/dftb-plus-user/attachments/20141221/1acb4761/attachment.html>