September 2012 - mpi-comments - mailman3.cels.anl.gov

Re: [Mpi-comments] "safe" tag in MPI_Intercomm_create
by William Gropp 07 Sep '12

07 Sep '12

Ok, I'll delete safe. 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 1:23 PM, Rolf Rabenseifner wrote: > 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(a)mcs.anl.gov> >> To: mpi-comments(a)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(a)illinois.edu> >>>>> To: "Comments on the MPI Standard" >>>>> <mpi-comments(a)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(a)lists.mpi-forum.org >>>>>>>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments >>>>>>> >>>>>>> >>>>>>> _______________________________________________ >>>>>>> mpi-comments mailing list >>>>>>> mpi-comments(a)lists.mpi-forum.org >>>>>>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments >>>>>>> >>>>>> _______________________________________________ >>>>>> mpi-comments mailing list >>>>>> mpi-comments(a)lists.mpi-forum.org >>>>>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments >>>>> >>>>> >>>>> _______________________________________________ >>>>> mpi-comments mailing list >>>>> mpi-comments(a)lists.mpi-forum.org >>>>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments >>>> >>>> -- >>>> Dr. Rolf Rabenseifner . . . . . . . . . .. email >>>> rabenseifner(a)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(a)lists.mpi-forum.org >>>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments >>> >>> _______________________________________________ >>> mpi-comments mailing list >>> mpi-comments(a)lists.mpi-forum.org >>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments >>> >> _______________________________________________ >> mpi-comments mailing list >> mpi-comments(a)lists.mpi-forum.org >> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments > > -- > Dr. Rolf Rabenseifner . . . . . . . . . .. email rabenseifner(a)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(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments

1 0

Re: [Mpi-comments] "safe" tag in MPI_Intercomm_create
by Rolf Rabenseifner 07 Sep '12

07 Sep '12

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(a)mcs.anl.gov> > To: mpi-comments(a)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(a)illinois.edu> > >>> To: "Comments on the MPI Standard" > >>> <mpi-comments(a)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(a)lists.mpi-forum.org > >>>>>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments > >>>>> > >>>>> > >>>>> _______________________________________________ > >>>>> mpi-comments mailing list > >>>>> mpi-comments(a)lists.mpi-forum.org > >>>>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments > >>>>> > >>>> _______________________________________________ > >>>> mpi-comments mailing list > >>>> mpi-comments(a)lists.mpi-forum.org > >>>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments > >>> > >>> > >>> _______________________________________________ > >>> mpi-comments mailing list > >>> mpi-comments(a)lists.mpi-forum.org > >>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments > >> > >> -- > >> Dr. Rolf Rabenseifner . . . . . . . . . .. email > >> rabenseifner(a)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(a)lists.mpi-forum.org > >> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments > > > > _______________________________________________ > > mpi-comments mailing list > > mpi-comments(a)lists.mpi-forum.org > > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments > > > _______________________________________________ > mpi-comments mailing list > mpi-comments(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments -- Dr. Rolf Rabenseifner . . . . . . . . . .. email rabenseifner(a)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)

1 0

Re: [Mpi-comments] "safe" tag in MPI_Intercomm_create
by Jim Dinan 07 Sep '12

07 Sep '12

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(a)illinois.edu> >>> To: "Comments on the MPI Standard" <mpi-comments(a)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(a)lists.mpi-forum.org >>>>>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments >>>>> >>>>> >>>>> _______________________________________________ >>>>> mpi-comments mailing list >>>>> mpi-comments(a)lists.mpi-forum.org >>>>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments >>>>> >>>> _______________________________________________ >>>> mpi-comments mailing list >>>> mpi-comments(a)lists.mpi-forum.org >>>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments >>> >>> >>> _______________________________________________ >>> mpi-comments mailing list >>> mpi-comments(a)lists.mpi-forum.org >>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments >> >> -- >> Dr. Rolf Rabenseifner . . . . . . . . . .. email rabenseifner(a)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(a)lists.mpi-forum.org >> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments > > _______________________________________________ > mpi-comments mailing list > mpi-comments(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments >

1 0

Re: [Mpi-comments] "safe" tag in MPI_Intercomm_create
by William Gropp 07 Sep '12

07 Sep '12

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(a)illinois.edu> >> To: "Comments on the MPI Standard" <mpi-comments(a)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(a)lists.mpi-forum.org >>>>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments >>>> >>>> >>>> _______________________________________________ >>>> mpi-comments mailing list >>>> mpi-comments(a)lists.mpi-forum.org >>>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments >>>> >>> _______________________________________________ >>> mpi-comments mailing list >>> mpi-comments(a)lists.mpi-forum.org >>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments >> >> >> _______________________________________________ >> mpi-comments mailing list >> mpi-comments(a)lists.mpi-forum.org >> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments > > -- > Dr. Rolf Rabenseifner . . . . . . . . . .. email rabenseifner(a)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(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments

1 0

Re: [Mpi-comments] "safe" tag in MPI_Intercomm_create
by Rolf Rabenseifner 07 Sep '12

07 Sep '12

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(a)illinois.edu> > To: "Comments on the MPI Standard" <mpi-comments(a)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(a)lists.mpi-forum.org > >>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments > >> > >> > >> _______________________________________________ > >> mpi-comments mailing list > >> mpi-comments(a)lists.mpi-forum.org > >> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments > >> > > _______________________________________________ > > mpi-comments mailing list > > mpi-comments(a)lists.mpi-forum.org > > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments > > > _______________________________________________ > mpi-comments mailing list > mpi-comments(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments -- Dr. Rolf Rabenseifner . . . . . . . . . .. email rabenseifner(a)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)

1 0

Re: [Mpi-comments] "safe" tag in MPI_Intercomm_create
by William Gropp 07 Sep '12

07 Sep '12

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(a)lists.mpi-forum.org >>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments >> >> >> _______________________________________________ >> mpi-comments mailing list >> mpi-comments(a)lists.mpi-forum.org >> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments >> > _______________________________________________ > mpi-comments mailing list > mpi-comments(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments

1 0

Re: [Mpi-comments] "safe" tag in MPI_Intercomm_create
by Jim Dinan 06 Sep '12

06 Sep '12

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(a)lists.mpi-forum.org >> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments > > > _______________________________________________ > mpi-comments mailing list > mpi-comments(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments >

1 0

Re: [Mpi-comments] "safe" tag in MPI_Intercomm_create
by William Gropp 06 Sep '12

06 Sep '12

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(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments

1 0

"safe" tag in MPI_Intercomm_create
by Jim Dinan 06 Sep '12

06 Sep '12

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.

1 0

Tags for Nonblocking Collective Operations
by Qingyu Meng 05 Sep '12

05 Sep '12

Draft URL: http://meetings.mpi-forum.org/mpi3.0_draft_2.pdf Page 196 5.12 Nonblocking Collective Operations The proposed MPI-3 non-blocking collective operations are designed for better overlapping computation and communication. The matching of these operations is based on the order of the calls instead of using tags. This limits the possibility to use dynamic code/task execution when the order of calls to collective operations should be decided at run-time. When running with multiple threads, our computational framework, Uintah (www.uintah.utah.edu) has to duplicate MPI communicators to ensure collective messages get to the right buffer. The current draft version of the MPI-3 standard states, ï¿½The use of tags for collective operations can prevent certain hardware optimizations.ï¿½ (Page. 108, Line. 14). However, the standard does allow the use of communicators to match collective operations. Does this usage also prevent certain hardware optimizations? Duplicating communicators for message matching is inefficient, as communicators must be reserved before use and also cause additional overhead. We would like to request further consideration in supporting the use of tags in nonblocking collective operations, to avoid the inefficiencies related to communicator duplication. This would not only benefit our computational framework, but other task-based, parallel frameworks that rely on collective operations as we move toward exascale. Qingyu Meng Scientific Computing and Imaging Institute, University of Utah

1 0