[Mpi-comments] "safe" tag in MPI_Intercomm_create

Rolf Rabenseifner rabenseifner at hlrs.de
Fri Sep 7 13:23:38 CDT 2012


Sounds good to me.
The tag is still usefull as a security argument:
If one process provides another tag (i.e., a wrong tag),
then this routine may or will produce a deadlock,
similar to pt-to-pt.

Rolf

----- Original Message -----
> From: "Jim Dinan" <dinan at mcs.anl.gov>
> To: mpi-comments at lists.mpi-forum.org
> Sent: Friday, September 7, 2012 8:14:29 PM
> Subject: Re: [Mpi-comments] "safe" tag in MPI_Intercomm_create
> I think that we should delete "safe" and leave any further changes for
> MPI-next.
> 
> ~Jim.
> 
> On 9/7/12 12:44 PM, William Gropp wrote:
> > Good point. This moves this to MPI-next.
> >
> > Bill
> >
> > William Gropp
> > Director, Parallel Computing Institute
> > Deputy Director for Research
> > Institute for Advanced Computing Applications and Technologies
> > Paul and Cynthia Saylor Professor of Computer Science
> > University of Illinois Urbana-Champaign
> >
> >
> >
> > On Sep 7, 2012, at 12:22 PM, Rolf Rabenseifner wrote:
> >
> >> If we ever make this routine a nonblocking collective,
> >> the tag may be needed again, because I expect that
> >> then, no two nonblocking MPI_Intercomm_create can be
> >> active with the same tag.
> >>
> >> This means, one may include backward and future use.
> >> It should not be a dummy. It should continue to
> >> require to have a valid tag, i.e., in the tag space
> >> between 0 and the value returned with inquiring MPI_TAG.
> >>
> >> Rolf
> >>
> >> ----- Original Message -----
> >>> From: "William Gropp" <wgropp at illinois.edu>
> >>> To: "Comments on the MPI Standard"
> >>> <mpi-comments at lists.mpi-forum.org>
> >>> Sent: Friday, September 7, 2012 5:55:14 PM
> >>> Subject: Re: [Mpi-comments] "safe" tag in MPI_Intercomm_create
> >>> Is this an MPI-next or a ticket zero change? Is this a chapter
> >>> committee change (I think it should be, but I don't want to make
> >>> that
> >>> call unilaterally)?
> >>>
> >>> Bill
> >>>
> >>> William Gropp
> >>> Director, Parallel Computing Institute
> >>> Deputy Director for Research
> >>> Institute for Advanced Computing Applications and Technologies
> >>> Paul and Cynthia Saylor Professor of Computer Science
> >>> University of Illinois Urbana-Champaign
> >>>
> >>>
> >>>
> >>> On Sep 6, 2012, at 5:36 PM, Jim Dinan wrote:
> >>>
> >>>> Hi Bill,
> >>>>
> >>>> That's right, the tag argument is vestigial as of MPI 3.0. I
> >>>> agree
> >>>> with your suggestion of changing '"safe" tag' to:
> >>>>
> >>>> dummy - Ignored argument, retained for backward compatibility.
> >>>>
> >>>> ~Jim.
> >>>>
> >>>> On 9/6/12 5:02 PM, William Gropp wrote:
> >>>>> I noted in ticket #305 that the argument is now ignored and
> >>>>> retained only for backward compatibility. Deleting "safe" isn't
> >>>>> enough.
> >>>>>
> >>>>> Bill
> >>>>>
> >>>>> William Gropp
> >>>>> Director, Parallel Computing Institute
> >>>>> Deputy Director for Research
> >>>>> Institute for Advanced Computing Applications and Technologies
> >>>>> Paul and Cynthia Saylor Professor of Computer Science
> >>>>> University of Illinois Urbana-Champaign
> >>>>>
> >>>>>
> >>>>>
> >>>>> On Sep 6, 2012, at 1:47 PM, Jim Dinan wrote:
> >>>>>
> >>>>>> Hi All,
> >>>>>>
> >>>>>> It looks like we forgot to include the removal of the "safe"
> >>>>>> qualifier from the tag argument as a part of the change to
> >>>>>> MPI_Intercomm_create:
> >>>>>>
> >>>>>> https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/305
> >>>>>>
> >>>>>> Specifying that the tag argument must be "safe" was made
> >>>>>> unnecessary by this change, since it eliminates the possibility
> >>>>>> of
> >>>>>> interference with user communication.
> >>>>>>
> >>>>>> ~Jim.
> >>>>>> _______________________________________________
> >>>>>> mpi-comments mailing list
> >>>>>> mpi-comments at lists.mpi-forum.org
> >>>>>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments
> >>>>>
> >>>>>
> >>>>> _______________________________________________
> >>>>> mpi-comments mailing list
> >>>>> mpi-comments at lists.mpi-forum.org
> >>>>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments
> >>>>>
> >>>> _______________________________________________
> >>>> mpi-comments mailing list
> >>>> mpi-comments at lists.mpi-forum.org
> >>>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments
> >>>
> >>>
> >>> _______________________________________________
> >>> mpi-comments mailing list
> >>> mpi-comments at lists.mpi-forum.org
> >>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments
> >>
> >> --
> >> Dr. Rolf Rabenseifner . . . . . . . . . .. email
> >> rabenseifner at hlrs.de
> >> High Performance Computing Center (HLRS) . phone
> >> ++49(0)711/685-65530
> >> University of Stuttgart . . . . . . . . .. fax ++49(0)711 /
> >> 685-65832
> >> Head of Dpmt Parallel Computing . . .
> >> www.hlrs.de/people/rabenseifner
> >> Nobelstr. 19, D-70550 Stuttgart, Germany . (Office: Allmandring 30)
> >> _______________________________________________
> >> mpi-comments mailing list
> >> mpi-comments at lists.mpi-forum.org
> >> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments
> >
> > _______________________________________________
> > mpi-comments mailing list
> > mpi-comments at lists.mpi-forum.org
> > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments
> >
> _______________________________________________
> mpi-comments mailing list
> mpi-comments at lists.mpi-forum.org
> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments

-- 
Dr. Rolf Rabenseifner . . . . . . . . . .. email rabenseifner at hlrs.de
High Performance Computing Center (HLRS) . phone ++49(0)711/685-65530
University of Stuttgart . . . . . . . . .. fax ++49(0)711 / 685-65832
Head of Dpmt Parallel Computing . . . www.hlrs.de/people/rabenseifner
Nobelstr. 19, D-70550 Stuttgart, Germany . (Office: Allmandring 30)



More information about the mpi-comments mailing list