Open MPI logo

Open MPI Development Mailing List Archives

  |   Home   |   Support   |   FAQ   |   all Development mailing list

Subject: Re: [OMPI devel] 1.7.4rc: MPI_F08_TYPE build failure with AMD's Open64
From: Paul Hargrove (phhargrove_at_[hidden])
Date: 2014-01-21 17:36:10


On Tue, Jan 21, 2014 at 1:55 PM, Jeff Squyres (jsquyres) <jsquyres_at_[hidden]
> wrote:

> Paul --
>
> I'm sorry, due to craziness and the holiday yesterday, the amended Fortran
> BIND(C) checks didn't get committed to the v1.7 branch until earlier today.
> So they'll be in tonight's tarball.
>
> It looks to me like the Open64 compilers won't pass the BIND(C) checks,
> and we should be ok. Can you verify tonight?
>

Will do. I owe Ralph a check on the MIPS64 atomics too.

>
> As for adding --disable-mpi-fortran-08... maybe. I'd still prefer the
> configure script to just Do The Right Thing. But I see your point for
> handling the one-off cases where it doesn't.
>

If ones compiler family of choice (e.g. XLF on PPC) passes the configure
checks but fails building the F08 stuff, then the presence or absence of
such a configuration option comes down to the difference between amputation
and euthanasia. With the option, one can surgically remove F08 support and
retain F77 and F90 support, but without it one must discard the fortran
compiler entirely. Since the likely alternative to a vendor compiler is
gfortran (which fails the F08 configure checks gracefully), you STILL don't
have F08 support and very possibly have slower executables (if you believe
that vendors compilers are better than GNU).

So, I think --disable-mpi-fortran-08 would be a valuable addition. Since
the configure logic is already fine-grained (OMPI_BUILD_FORTAN_*_BINDINGS),
it shouldn't be a big deal to bring back the former --disable switches for
F77 and F90 support, though I am not requesting that you do so.

-Paul

>
>
> On Jan 17, 2014, at 7:49 PM, Paul Hargrove <phhargrove_at_[hidden]> wrote:
>
> > Building the v1.7 tarball (1.7.4rc2r30303) with AMD's Open64 compilers
> (v4.5.2) I hit the errors shown in the attached make.log (too long to
> cut-and-paste).
> > I've also attached config.log and configure's stdout.
> >
> > "We don't care about that compiler" is an acceptable (to me) answer, but
> I am reporting this for completeness.
> >
> > Jeff may need to do more work on the configure probes for F08 support.
> > OR, why not provide --disable-mpi-f08 ?
> >
> > -Paul
> >
> > --
> > Paul H. Hargrove PHHargrove_at_[hidden]
> > Future Technologies Group
> > Computer and Data Sciences Department Tel: +1-510-495-2352
> > Lawrence Berkeley National Laboratory Fax: +1-510-486-6900
> >
> <configure.stdout.bz2><config.log.bz2><make.log.bz2>_______________________________________________
> > devel mailing list
> > devel_at_[hidden]
> > http://www.open-mpi.org/mailman/listinfo.cgi/devel
>
>
> --
> Jeff Squyres
> jsquyres_at_[hidden]
> For corporate legal information go to:
> http://www.cisco.com/web/about/doing_business/legal/cri/
>
> _______________________________________________
> devel mailing list
> devel_at_[hidden]
> http://www.open-mpi.org/mailman/listinfo.cgi/devel
>

-- 
Paul H. Hargrove                          PHHargrove_at_[hidden]
Future Technologies Group
Computer and Data Sciences Department     Tel: +1-510-495-2352
Lawrence Berkeley National Laboratory     Fax: +1-510-486-6900