[MPI3 Fortran] Deprecate mpif.h?
Bill Long
longb at cray.com
Wed Mar 3 17:54:20 CST 2010
Supalov, Alexander wrote:
> Thanks. Let me sketch a sad yet very realistic artificial example.
>
> Imagine a hapless owner of 10-million lines of mostly Fortran IV code, with some C inlays, whose original authors retired 25 years ago. Every single change after that goes thru the ISO 9000 certification process. And still they need to get the 5% performance increase the NB collectives may bring.
>
> Now tell them they must "use mpi3" for that. Shudder.
I'm sure there are several example like this. But, to use the NB
collectives, they will need to modify the code, right? If so, then they
will have to go through recertification anyway, so adding the use mpi3
line (and deleting the include mpif.h line) would add only tiny amounts
of extra pain (99% of which is in the recertification). Or did I miss
something here?
Cheers,
Bill
>
> -----Original Message-----
> From: mpi3-fortran-bounces at lists.mpi-forum.org [mailto:mpi3-fortran-bounces at lists.mpi-forum.org] On Behalf Of Jeff Squyres
> Sent: Wednesday, March 03, 2010 10:22 PM
> To: MPI-3 Fortran working group
> Subject: Re: [MPI3 Fortran] Deprecate mpif.h?
>
> On Mar 3, 2010, at 4:06 PM, Supalov, Alexander wrote:
>
>> I can imagine a venerable application that wants to retain all of F77 stuff but use the nonblocking collectives, for example. Would this be considered a "minor" mpif.h modification?
>
> Not if it's deprecated. If it's deprecated, nothing new will be included in mpif.h.
>
> The path forward for such an app would be to use mpi3 for the parts of their app that want to use nonblocking collectives. That was the explicit goal of having INTEGER<-->Fortran handle conversion functions (i.e., legacy apps that include mpif.h/use mpi, but want to add some MPI-3 functionality).
>
--
Bill Long longb at cray.com
Fortran Technical Support & voice: 651-605-9024
Bioinformatics Software Development fax: 651-605-9142
Cray Inc./Cray Plaza, Suite 210/380 Jackson St./St. Paul, MN 55101
More information about the mpiwg-fortran
mailing list