September 2012 - mpiwg-rma - mailman3.cels.anl.gov

[Mpi3-rma] Urgent: Dynamic window meeting
by Jim Dinan 10 Sep '12

10 Sep '12

Hi All, The editing period for MPI-3 ends tomorrow and we still need to reach a resolution on the displacements in dynamic windows issue. I'm not sure which means of communication will be most effective. We can try email, but in the meantime, I also created a doodle poll to meet tomorrow. Please fill in your availability: http://www.doodle.com/2kkbynwarkpyqkfe Thanks, ~Jim.

1 0

Re: [Mpi3-rma] Fwd: [Mpi-forum] MPI 3.0 overview
by maik peterson 10 Sep '12

10 Sep '12

please dont forget to add some information about mpi/pgas coexistence today and in the future. without that more and more people will migrate away from mpi. maik 2012/9/7 William Gropp <wgropp(a)illinois.edu> > One of us needs to provide this. There are the slides that we've used for > the forum itself, and abbreviated versions of these. I'm working on the > standard updates - is someone able to prepare these slides for Rich? > > 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 > > > > Begin forwarded message: > > *From: *Richard Graham <richardg(a)mellanox.com> > *Subject: **[Mpi-forum] MPI 3.0 overview* > *Date: *September 4, 2012 2:17:29 AM GMT+02:00 > *To: *"mpi-forum(a)lists.mpi-forum.org" <mpi-forum(a)lists.mpi-forum.org> > *Reply-To: *Main MPI Forum mailing list <mpi-forum(a)lists.mpi-forum.org> > > In preparation for presentations on the MPI 3.0 standard, I would like to > ask that the working groups send a few slides on the following new > additions to the MPI standard:**** > ** ** > - Nonblocking collectives**** > - RMA enhancements**** > - MPI Tool Interface**** > - New Fortran bindings**** > - Sparse collectives**** > - Enhanced Datatype support**** > - Distributed Graph**** > - Noncollective communicator creation**** > ** ** > Also, if your working group had items that did not make it into MPI-3.0, > and would like these mentioned in the same talk, please provide me with > some slides. It is good to let folks know why some items did not make it > into the standard.**** > ** ** > It would be very helpful if I could get these slides by Sept the 12th.**** > ** ** > Thanks,**** > Rich**** > _______________________________________________ > mpi-forum mailing list > mpi-forum(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum > > > > _______________________________________________ > mpi3-rma mailing list > mpi3-rma(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-rma >

1 0

Re: [Mpi3-rma] [Mpi-forum] MPI 3.0 overview
by William Gropp 08 Sep '12

08 Sep '12

1 slide would be helpful (Rich has a lot of material to cover, so we need to be concise). Thanks! 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 4:08 PM, Jeff Hammond wrote: > I can provide a slide or two on my pet, request-based local completion > RMA, and why it is _enormously useful_ :-) if encouraged to do so. > > It might make sense for someone to say something about MPI-3 in > relation to PGAS, SHMEM and/or ARMCI, since at least some fraction of > the one-sided people in the world are going to measure us in that way. > > Jeff > > On Fri, Sep 7, 2012 at 1:26 PM, Jim Dinan <dinan(a)mcs.anl.gov> wrote: >> Hi Bill, >> >> I've attached a few MPI-3 RMA summary slides that Dave and I presented at a >> recent tutorial. We could use these as a starting point. >> >> ~Jim. >> >> >> On 9/7/12 2:52 PM, William Gropp wrote: >>> >>> One of us needs to provide this. There are the slides that we've used >>> for the forum itself, and abbreviated versions of these. I'm working on >>> the standard updates - is someone able to prepare these slides for Rich? >>> >>> 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 >>> >>> >>> >>> Begin forwarded message: >>> >>>> *From: *Richard Graham <richardg(a)mellanox.com >>>> <mailto:[email protected]>> >>>> *Subject: **[Mpi-forum] MPI 3.0 overview* >>>> *Date: *September 4, 2012 2:17:29 AM GMT+02:00 >>>> *To: *"mpi-forum(a)lists.mpi-forum.org >>>> <mailto:[email protected]>" <mpi-forum(a)lists.mpi-forum.org >>>> <mailto:[email protected]>> >>>> *Reply-To: *Main MPI Forum mailing list <mpi-forum(a)lists.mpi-forum.org >>>> <mailto:[email protected]>> >>>> >>>> >>>> In preparation for presentations on the MPI 3.0 standard, I would like >>>> to ask that the working groups send a few slides on the following new >>>> additions to the MPI standard: >>>> - Nonblocking collectives >>>> - RMA enhancements >>>> - MPI Tool Interface >>>> - New Fortran bindings >>>> - Sparse collectives >>>> - Enhanced Datatype support >>>> - Distributed Graph >>>> - Noncollective communicator creation >>>> Also, if your working group had items that did not make it into >>>> MPI-3.0, and would like these mentioned in the same talk, please >>>> provide me with some slides. It is good to let folks know why some >>>> items did not make it into the standard. >>>> It would be very helpful if I could get these slides by Sept the 12^th . >>>> >>>> Thanks, >>>> Rich >>>> _______________________________________________ >>>> mpi-forum mailing list >>>> mpi-forum(a)lists.mpi-forum.org <mailto:[email protected]> >>>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum >>> >>> >>> >>> >>> _______________________________________________ >>> mpi3-rma mailing list >>> mpi3-rma(a)lists.mpi-forum.org >>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-rma >>> >> >> _______________________________________________ >> mpi3-rma mailing list >> mpi3-rma(a)lists.mpi-forum.org >> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-rma > > > > -- > Jeff Hammond > Argonne Leadership Computing Facility > University of Chicago Computation Institute > jhammond(a)alcf.anl.gov / (630) 252-5381 > http://www.linkedin.com/in/jeffhammond > https://wiki.alcf.anl.gov/parts/index.php/User:Jhammond > _______________________________________________ > mpi3-rma mailing list > mpi3-rma(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-rma

1 0

Re: [Mpi3-rma] MPI_NO_OP and origin buffer
by Jim Dinan 07 Sep '12

07 Sep '12

MPI_Accumulate with MPI_REPLACE accomplishes a put. NO_OP is not permitted in MPI_Accumulate. On 9/7/12 5:23 PM, Torsten Hoefler wrote: > All, > > So it may be fine for some accumulate ops but not for all because > MPI_Accumulate with MPI_NO_OP was our back-door to get an atomic put. > Thanks to Keith for pointing this out! > > Thanks, > Torsten > > On Fri, Sep 07, 2012 at 04:40:54PM -0500, Torsten Hoefler wrote: >> Hi Jim, >> >> I agree, that is a reasonable change. >> >> Best, >> Torsten >> >> On Thu, Sep 06, 2012 at 10:48:47AM -0500, James Dinan wrote: >>> Hi All, >>> >>> It looks like we forgot to specify that for MPI_NO_OP in the new >>> accumulate ops, the origin buffer is ignored (i.e. it's acceptable to >>> pass NULL). We can add this text on page 441, line 30: >>> >>> When MPI_NO_OP is specified, the origin buffer is ignored. >>> >>> We already have such a use case in the linked list example. >>> >>> ~Jim. >>> _______________________________________________ >>> mpi3-rma mailing list >>> mpi3-rma(a)lists.mpi-forum.org >>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-rma >>> >> >> -- >> ### qreharg rug ebs fv crryF ------------- http://www.unixer.de/ ----- >> Torsten Hoefler | Assistant Professor >> Dept. of Computer Science | ETH Zï¿½rich >> Universitï¿½tsstrasse 6 | Zurich-8092, Switzerland >> CAB E 64.1 | Phone: +41 76 309 79 29 >> _______________________________________________ >> mpi3-rma mailing list >> mpi3-rma(a)lists.mpi-forum.org >> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-rma >> >

1 0

Re: [Mpi3-rma] MPI_NO_OP and origin buffer
by Torsten Hoefler 07 Sep '12

07 Sep '12

All, So it may be fine for some accumulate ops but not for all because MPI_Accumulate with MPI_NO_OP was our back-door to get an atomic put. Thanks to Keith for pointing this out! Thanks, Torsten On Fri, Sep 07, 2012 at 04:40:54PM -0500, Torsten Hoefler wrote: > Hi Jim, > > I agree, that is a reasonable change. > > Best, > Torsten > > On Thu, Sep 06, 2012 at 10:48:47AM -0500, James Dinan wrote: > > Hi All, > > > > It looks like we forgot to specify that for MPI_NO_OP in the new > > accumulate ops, the origin buffer is ignored (i.e. it's acceptable to > > pass NULL). We can add this text on page 441, line 30: > > > > When MPI_NO_OP is specified, the origin buffer is ignored. > > > > We already have such a use case in the linked list example. > > > > ~Jim. > > _______________________________________________ > > mpi3-rma mailing list > > mpi3-rma(a)lists.mpi-forum.org > > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-rma > > > > -- > ### qreharg rug ebs fv crryF ------------- http://www.unixer.de/ ----- > Torsten Hoefler | Assistant Professor > Dept. of Computer Science | ETH Zï¿½rich > Universitï¿½tsstrasse 6 | Zurich-8092, Switzerland > CAB E 64.1 | Phone: +41 76 309 79 29 > _______________________________________________ > mpi3-rma mailing list > mpi3-rma(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-rma > -- ### qreharg rug ebs fv crryF ------------- http://www.unixer.de/ ----- Torsten Hoefler | Assistant Professor Dept. of Computer Science | ETH Zï¿½rich Universitï¿½tsstrasse 6 | Zurich-8092, Switzerland CAB E 64.1 | Phone: +41 76 309 79 29

1 0

Re: [Mpi3-rma] MPI_NO_OP and origin buffer
by Jeff Hammond 07 Sep '12

07 Sep '12

if we are voting, +1. On Fri, Sep 7, 2012 at 2:40 PM, Torsten Hoefler <htor(a)illinois.edu> wrote: > Hi Jim, > > I agree, that is a reasonable change. > > Best, > Torsten > > On Thu, Sep 06, 2012 at 10:48:47AM -0500, James Dinan wrote: >> Hi All, >> >> It looks like we forgot to specify that for MPI_NO_OP in the new >> accumulate ops, the origin buffer is ignored (i.e. it's acceptable to >> pass NULL). We can add this text on page 441, line 30: >> >> When MPI_NO_OP is specified, the origin buffer is ignored. >> >> We already have such a use case in the linked list example. >> >> ~Jim. >> _______________________________________________ >> mpi3-rma mailing list >> mpi3-rma(a)lists.mpi-forum.org >> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-rma >> > > -- > ### qreharg rug ebs fv crryF ------------- http://www.unixer.de/ ----- > Torsten Hoefler | Assistant Professor > Dept. of Computer Science | ETH Zï¿½rich > Universitï¿½tsstrasse 6 | Zurich-8092, Switzerland > CAB E 64.1 | Phone: +41 76 309 79 29 > _______________________________________________ > mpi3-rma mailing list > mpi3-rma(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-rma -- Jeff Hammond Argonne Leadership Computing Facility University of Chicago Computation Institute jhammond(a)alcf.anl.gov / (630) 252-5381 http://www.linkedin.com/in/jeffhammond https://wiki.alcf.anl.gov/parts/index.php/User:Jhammond

1 0

Re: [Mpi3-rma] MPI_NO_OP and origin buffer
by Torsten Hoefler 07 Sep '12

07 Sep '12

Hi Jim, I agree, that is a reasonable change. Best, Torsten On Thu, Sep 06, 2012 at 10:48:47AM -0500, James Dinan wrote: > Hi All, > > It looks like we forgot to specify that for MPI_NO_OP in the new > accumulate ops, the origin buffer is ignored (i.e. it's acceptable to > pass NULL). We can add this text on page 441, line 30: > > When MPI_NO_OP is specified, the origin buffer is ignored. > > We already have such a use case in the linked list example. > > ~Jim. > _______________________________________________ > mpi3-rma mailing list > mpi3-rma(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-rma > -- ### qreharg rug ebs fv crryF ------------- http://www.unixer.de/ ----- Torsten Hoefler | Assistant Professor Dept. of Computer Science | ETH Zï¿½rich Universitï¿½tsstrasse 6 | Zurich-8092, Switzerland CAB E 64.1 | Phone: +41 76 309 79 29

1 0

Re: [Mpi3-rma] Fwd: [Mpi-forum] MPI 3.0 overview
by Jeff Hammond 07 Sep '12

07 Sep '12

I can provide a slide or two on my pet, request-based local completion RMA, and why it is _enormously useful_ :-) if encouraged to do so. It might make sense for someone to say something about MPI-3 in relation to PGAS, SHMEM and/or ARMCI, since at least some fraction of the one-sided people in the world are going to measure us in that way. Jeff On Fri, Sep 7, 2012 at 1:26 PM, Jim Dinan <dinan(a)mcs.anl.gov> wrote: > Hi Bill, > > I've attached a few MPI-3 RMA summary slides that Dave and I presented at a > recent tutorial. We could use these as a starting point. > > ~Jim. > > > On 9/7/12 2:52 PM, William Gropp wrote: >> >> One of us needs to provide this. There are the slides that we've used >> for the forum itself, and abbreviated versions of these. I'm working on >> the standard updates - is someone able to prepare these slides for Rich? >> >> 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 >> >> >> >> Begin forwarded message: >> >>> *From: *Richard Graham <richardg(a)mellanox.com >>> <mailto:[email protected]>> >>> *Subject: **[Mpi-forum] MPI 3.0 overview* >>> *Date: *September 4, 2012 2:17:29 AM GMT+02:00 >>> *To: *"mpi-forum(a)lists.mpi-forum.org >>> <mailto:[email protected]>" <mpi-forum(a)lists.mpi-forum.org >>> <mailto:[email protected]>> >>> *Reply-To: *Main MPI Forum mailing list <mpi-forum(a)lists.mpi-forum.org >>> <mailto:[email protected]>> >>> >>> >>> In preparation for presentations on the MPI 3.0 standard, I would like >>> to ask that the working groups send a few slides on the following new >>> additions to the MPI standard: >>> - Nonblocking collectives >>> - RMA enhancements >>> - MPI Tool Interface >>> - New Fortran bindings >>> - Sparse collectives >>> - Enhanced Datatype support >>> - Distributed Graph >>> - Noncollective communicator creation >>> Also, if your working group had items that did not make it into >>> MPI-3.0, and would like these mentioned in the same talk, please >>> provide me with some slides. It is good to let folks know why some >>> items did not make it into the standard. >>> It would be very helpful if I could get these slides by Sept the 12^th . >>> >>> Thanks, >>> Rich >>> _______________________________________________ >>> mpi-forum mailing list >>> mpi-forum(a)lists.mpi-forum.org <mailto:[email protected]> >>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum >> >> >> >> >> _______________________________________________ >> mpi3-rma mailing list >> mpi3-rma(a)lists.mpi-forum.org >> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-rma >> > > _______________________________________________ > mpi3-rma mailing list > mpi3-rma(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-rma -- Jeff Hammond Argonne Leadership Computing Facility University of Chicago Computation Institute jhammond(a)alcf.anl.gov / (630) 252-5381 http://www.linkedin.com/in/jeffhammond https://wiki.alcf.anl.gov/parts/index.php/User:Jhammond

1 0

Re: [Mpi3-rma] Fwd: [Mpi-forum] MPI 3.0 overview
by Jim Dinan 07 Sep '12

07 Sep '12

Hi Bill, I've attached a few MPI-3 RMA summary slides that Dave and I presented at a recent tutorial. We could use these as a starting point. ~Jim. On 9/7/12 2:52 PM, William Gropp wrote: > One of us needs to provide this. There are the slides that we've used > for the forum itself, and abbreviated versions of these. I'm working on > the standard updates - is someone able to prepare these slides for Rich? > > 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 > > > > Begin forwarded message: > >> *From: *Richard Graham <richardg(a)mellanox.com >> <mailto:[email protected]>> >> *Subject: **[Mpi-forum] MPI 3.0 overview* >> *Date: *September 4, 2012 2:17:29 AM GMT+02:00 >> *To: *"mpi-forum(a)lists.mpi-forum.org >> <mailto:[email protected]>" <mpi-forum(a)lists.mpi-forum.org >> <mailto:[email protected]>> >> *Reply-To: *Main MPI Forum mailing list <mpi-forum(a)lists.mpi-forum.org >> <mailto:[email protected]>> >> >> In preparation for presentations on the MPI 3.0 standard, I would like >> to ask that the working groups send a few slides on the following new >> additions to the MPI standard: >> - Nonblocking collectives >> - RMA enhancements >> - MPI Tool Interface >> - New Fortran bindings >> - Sparse collectives >> - Enhanced Datatype support >> - Distributed Graph >> - Noncollective communicator creation >> Also, if your working group had items that did not make it into >> MPI-3.0, and would like these mentioned in the same talk, please >> provide me with some slides. It is good to let folks know why some >> items did not make it into the standard. >> It would be very helpful if I could get these slides by Sept the 12^th . >> Thanks, >> Rich >> _______________________________________________ >> mpi-forum mailing list >> mpi-forum(a)lists.mpi-forum.org <mailto:[email protected]> >> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum > > > > _______________________________________________ > mpi3-rma mailing list > mpi3-rma(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-rma >

1 0

[Mpi3-rma] Fwd: [Mpi-forum] MPI 3.0 overview
by William Gropp 07 Sep '12

07 Sep '12

One of us needs to provide this. There are the slides that we've used for the forum itself, and abbreviated versions of these. I'm working on the standard updates - is someone able to prepare these slides for Rich? 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 Begin forwarded message: > From: Richard Graham <richardg(a)mellanox.com> > Subject: [Mpi-forum] MPI 3.0 overview > Date: September 4, 2012 2:17:29 AM GMT+02:00 > To: "mpi-forum(a)lists.mpi-forum.org" <mpi-forum(a)lists.mpi-forum.org> > Reply-To: Main MPI Forum mailing list <mpi-forum(a)lists.mpi-forum.org> > > In preparation for presentations on the MPI 3.0 standard, I would like to ask that the working groups send a few slides on the following new additions to the MPI standard: > > - Nonblocking collectives > - RMA enhancements > - MPI Tool Interface > - New Fortran bindings > - Sparse collectives > - Enhanced Datatype support > - Distributed Graph > - Noncollective communicator creation > > Also, if your working group had items that did not make it into MPI-3.0, and would like these mentioned in the same talk, please provide me with some slides. It is good to let folks know why some items did not make it into the standard. > > It would be very helpful if I could get these slides by Sept the 12th. > > Thanks, > Rich > _______________________________________________ > mpi-forum mailing list > mpi-forum(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum

1 0