- Mpi-22 - mailman3.cels.anl.gov

Re: [Mpi-22] Ticket #33 (Scalable Graph Topology Interface)
by unknown＠example.com 21 Jul '09

21 Jul '09

Thanks for all your efforts and persistence, Torsten. On Jul 17, 2009, at 12:52 PM, Torsten Hoefler wrote: > Hello, > this is a friendly reminder for everybody with doubts to review ticket > #33. This is a complex ticket and I would like to ensure thorough > reviews. > > There is only one week left and I would like everybody, who thinks > that > the ticket is not good for MPI-2.2, incorrect, or has any other issues > with it, to speak up (on the mailinglist or via personal mail) > before the > meeting. > > All the positive effects of #33 are summarized in the ticket and in my > last mail (http://lists.mpi-forum.org/mpi-22/2009/06/0454.php) > > I will also try to reserve some time before the final vote to discuss > last doubts. I am not trying to rush the ticket, I want MPI-2.2 to be > good and meaningful. I will happily postpone the fix to MPI-3 if there > are good arguments against the ticket. I addressed all criticisms > about > the non-scalable implementation and don't see any problems. Please > review it carefully! > > Thanks & All the Best, > Torsten > > -- > bash$ :(){ :|:&};: --------------------- http://www.unixer.de/ ----- > Torsten Hoefler, Ph.D. | Postdoctoral Fellow > Open Systems Lab | Indiana University > 150 S. Woodlawn Ave. | Bloomington, IN, 474045, USA > Lindley Hall Room 135 | +01 (812) 855-3608 > _______________________________________________ > mpi-22 mailing list > mpi-22_at_[hidden] > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-22 > -- Jeff Squyres jsquyres_at_[hidden]

1 0

Re: [Mpi-22] Ticket #33 (Scalable Graph Topology Interface)
by unknown＠example.com 17 Jul '09

17 Jul '09

Hello, this is a friendly reminder for everybody with doubts to review ticket #33. This is a complex ticket and I would like to ensure thorough reviews. There is only one week left and I would like everybody, who thinks that the ticket is not good for MPI-2.2, incorrect, or has any other issues with it, to speak up (on the mailinglist or via personal mail) before the meeting. All the positive effects of #33 are summarized in the ticket and in my last mail (http://lists.mpi-forum.org/mpi-22/2009/06/0454.php) I will also try to reserve some time before the final vote to discuss last doubts. I am not trying to rush the ticket, I want MPI-2.2 to be good and meaningful. I will happily postpone the fix to MPI-3 if there are good arguments against the ticket. I addressed all criticisms about the non-scalable implementation and don't see any problems. Please review it carefully! Thanks & All the Best, Torsten -- bash$ :(){ :|:&};: --------------------- http://www.unixer.de/ ----- Torsten Hoefler, Ph.D. | Postdoctoral Fellow Open Systems Lab | Indiana University 150 S. Woodlawn Ave. | Bloomington, IN, 474045, USA Lindley Hall Room 135 | +01 (812) 855-3608

1 0

Re: [Mpi-22] Ticket #33 (Scalable Graph Topology Interface)
by unknown＠example.com 19 Jun '09

19 Jun '09

I agree with Bill - I think we should continue to extend the C++ bindings with new interfaces until at least 3.0. I think it would be very weird for users if a feature was available in C and Fortran and not C++. That being said, I also think the whole ABI thing is a nice goal, but not one of the original requirements of the 2.x process, so if something could maybe on some platform break the ABI compatibility for the C++ bindings, that should not be disqualifying for the 2.2 process. Brian On 6/19/09 9:19 , "William Gropp" <wgropp_at_[hidden]> wrote: > Deprecated doesn't mean removed - we will need some C++ binding for > the new features. > > Bill > > On Jun 19, 2009, at 5:31 AM, Jeff Squyres wrote: > >> On Jun 19, 2009, at 5:12 AM, Supalov, Alexander wrote: >> >>> Thanks. If we decide to deprecate C++ binding, will it still be >>> updated in the standard to expose the new MPI-2.2 features, or will >>> it be frozen at the MPI-2.1 level? >>> >> >> I'd be in favor of freezing it at 2.1 level -- then many of these ABI >> issues go away. >> >> -- >> Jeff Squyres >> Cisco Systems >> >> _______________________________________________ >> mpi-22 mailing list >> mpi-22_at_[hidden] >> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-22 > > William Gropp > Deputy Director for Research > Institute for Advanced Computing Applications and Technologies > Paul and Cynthia Saylor Professor of Computer Science > University of Illinois Urbana-Champaign > > > > > _______________________________________________ > mpi-22 mailing list > mpi-22_at_[hidden] > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-22 > -- Brian W. Barrett Dept. 1423: Scalable System Software Sandia National Laboratories

1 0

Re: [Mpi-22] Ticket #33 (Scalable Graph Topology Interface)
by unknown＠example.com 19 Jun '09

19 Jun '09

Deprecated doesn't mean removed - we will need some C++ binding for the new features. Bill On Jun 19, 2009, at 5:31 AM, Jeff Squyres wrote: > On Jun 19, 2009, at 5:12 AM, Supalov, Alexander wrote: > >> Thanks. If we decide to deprecate C++ binding, will it still be >> updated in the standard to expose the new MPI-2.2 features, or will >> it be frozen at the MPI-2.1 level? >> > > I'd be in favor of freezing it at 2.1 level -- then many of these ABI > issues go away. > > -- > Jeff Squyres > Cisco Systems > > _______________________________________________ > mpi-22 mailing list > mpi-22_at_[hidden] > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-22 William Gropp Deputy Director for Research Institute for Advanced Computing Applications and Technologies Paul and Cynthia Saylor Professor of Computer Science University of Illinois Urbana-Champaign

1 0

Re: [Mpi-22] Ticket #33 (Scalable Graph Topology Interface)
by unknown＠example.com 19 Jun '09

19 Jun '09

On Jun 19, 2009, at 5:12 AM, Supalov, Alexander wrote: > Thanks. If we decide to deprecate C++ binding, will it still be > updated in the standard to expose the new MPI-2.2 features, or will > it be frozen at the MPI-2.1 level? > I'd be in favor of freezing it at 2.1 level -- then many of these ABI issues go away. -- Jeff Squyres Cisco Systems

1 0

Re: [Mpi-22] Ticket #33 (Scalable Graph Topology Interface)
by unknown＠example.com 19 Jun '09

19 Jun '09

Thanks. If we decide to deprecate C++ binding, will it still be updated in the standard to expose the new MPI-2.2 features, or will it be frozen at the MPI-2.1 level? -----Original Message----- From: mpi-22-bounces_at_[hidden] [mailto:mpi-22-bounces_at_[hidden]] On Behalf Of William Gropp Sent: Thursday, June 18, 2009 11:29 PM To: MPI 2.2 Subject: Re: [Mpi-22] Ticket #33 (Scalable Graph Topology Interface) The ABI issue is one that is important but was not part of the original requirements (I went back to the original slides kicking off the 2.2 effort). It certainly is something to consider, and I agree that it is best to avoid ABI changes if possible. And this might be something that we should have specified at the beginning of the 2.2 (or 2.1) process, but we didn't. Fortunately, for Fortran and C (but not C++), the other requirements should make it possible for an implementation to meet this requirement. Bill On Jun 18, 2009, at 1:50 PM, Supalov, Alexander wrote: > Hi, > > Thanks. I'm not sure what part of the Wiki is meant when the > guidelines are mentioned. Is it "Some guidelines for submissions" at > the top of https://svn.mpi-forum.org/trac/mpi-forum-web/wiki/MpiTwoTwoWikiPage > ? > > If so, I'm can't see ABI compatibility there yet, only backward > compatibility in the sense that "it must not break existing correct > programs" that may be understood as "recompile and enjoy". However, > a forced recompilation is not what some customers will enjoy. > > Something mode definite, like "The intention of the Forum is that > MPI-2.2 C and Fortran interfaces stay binary compatible with their > MPI-2.1 counterparts" would probably suffice to address this. Note > that C++ is intentionally omitted here. > > Best regards. > > Alexander > > -----Original Message----- > From: mpi-22-bounces_at_[hidden] [mailto:mpi-22-bounces_at_[hidden] > ] On Behalf Of Torsten Hoefler > Sent: Thursday, June 18, 2009 8:18 PM > To: MPI 2.2 > Subject: Re: [Mpi-22] Ticket #33 (Scalable Graph Topology Interface) > > On Thu, Jun 18, 2009 at 12:52:48PM -0500, William Gropp wrote: >> I've added my recollection of our various discussions on this to the >> wiki. These are the ones that I've applied as chair of the MPI 2.2 >> effort. > Thank you very much for this clarification Bill! I think/hope we all > agree on this. > > Ticket #33 fulfills guideline 3. I also attached a BSD-like (that's > what > the lawyers at IU said) license file to all my implementations (#24, > #31, #33, #94). > > Could everybody who is concerned with licensing issues (Alexander, > Erez) > please check the license and tell me if this is ok for you? > > Everybody else, please look carefully at ticket #33 and direct any > questions and/or critics to me. > > Thanks! > Torsten > > -- > bash$ :(){ :|:&};: --------------------- http://www.unixer.de/ ----- > Torsten Hoefler, Ph.D. | Postdoctoral Fellow > Open Systems Lab | Indiana University > 150 S. Woodlawn Ave. | Bloomington, IN, 474045, USA > Lindley Hall Room 135 | +01 (812) 855-3608 > _______________________________________________ > mpi-22 mailing list > mpi-22_at_[hidden] > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-22 > --------------------------------------------------------------------- > Intel GmbH > Dornacher Strasse 1 > 85622 Feldkirchen/Muenchen Germany > Sitz der Gesellschaft: Feldkirchen bei Muenchen > Geschaeftsfuehrer: Douglas Lusk, Peter Gleissner, Hannes Schwaderer > Registergericht: Muenchen HRB 47456 Ust.-IdNr. > VAT Registration No.: DE129385895 > Citibank Frankfurt (BLZ 502 109 00) 600119052 > > This e-mail and any attachments may contain confidential material for > the sole use of the intended recipient(s). Any review or distribution > by others is strictly prohibited. If you are not the intended > recipient, please contact the sender and delete all copies. > > > _______________________________________________ > mpi-22 mailing list > mpi-22_at_[hidden] > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-22 William Gropp Deputy Director for Research Institute for Advanced Computing Applications and Technologies Paul and Cynthia Saylor Professor of Computer Science University of Illinois Urbana-Champaign _______________________________________________ mpi-22 mailing list mpi-22_at_[hidden] http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-22 --------------------------------------------------------------------- Intel GmbH Dornacher Strasse 1 85622 Feldkirchen/Muenchen Germany Sitz der Gesellschaft: Feldkirchen bei Muenchen Geschaeftsfuehrer: Douglas Lusk, Peter Gleissner, Hannes Schwaderer Registergericht: Muenchen HRB 47456 Ust.-IdNr. VAT Registration No.: DE129385895 Citibank Frankfurt (BLZ 502 109 00) 600119052 This e-mail and any attachments may contain confidential material for the sole use of the intended recipient(s). Any review or distribution by others is strictly prohibited. If you are not the intended recipient, please contact the sender and delete all copies.

1 0

Re: [Mpi-22] Ticket #33 (Scalable Graph Topology Interface)
by unknown＠example.com 18 Jun '09

18 Jun '09

The ABI issue is one that is important but was not part of the original requirements (I went back to the original slides kicking off the 2.2 effort). It certainly is something to consider, and I agree that it is best to avoid ABI changes if possible. And this might be something that we should have specified at the beginning of the 2.2 (or 2.1) process, but we didn't. Fortunately, for Fortran and C (but not C++), the other requirements should make it possible for an implementation to meet this requirement. Bill On Jun 18, 2009, at 1:50 PM, Supalov, Alexander wrote: > Hi, > > Thanks. I'm not sure what part of the Wiki is meant when the > guidelines are mentioned. Is it "Some guidelines for submissions" at > the top of https://svn.mpi-forum.org/trac/mpi-forum-web/wiki/MpiTwoTwoWikiPage > ? > > If so, I'm can't see ABI compatibility there yet, only backward > compatibility in the sense that "it must not break existing correct > programs" that may be understood as "recompile and enjoy". However, > a forced recompilation is not what some customers will enjoy. > > Something mode definite, like "The intention of the Forum is that > MPI-2.2 C and Fortran interfaces stay binary compatible with their > MPI-2.1 counterparts" would probably suffice to address this. Note > that C++ is intentionally omitted here. > > Best regards. > > Alexander > > -----Original Message----- > From: mpi-22-bounces_at_[hidden] [mailto:mpi-22-bounces_at_[hidden] > ] On Behalf Of Torsten Hoefler > Sent: Thursday, June 18, 2009 8:18 PM > To: MPI 2.2 > Subject: Re: [Mpi-22] Ticket #33 (Scalable Graph Topology Interface) > > On Thu, Jun 18, 2009 at 12:52:48PM -0500, William Gropp wrote: >> I've added my recollection of our various discussions on this to the >> wiki. These are the ones that I've applied as chair of the MPI 2.2 >> effort. > Thank you very much for this clarification Bill! I think/hope we all > agree on this. > > Ticket #33 fulfills guideline 3. I also attached a BSD-like (that's > what > the lawyers at IU said) license file to all my implementations (#24, > #31, #33, #94). > > Could everybody who is concerned with licensing issues (Alexander, > Erez) > please check the license and tell me if this is ok for you? > > Everybody else, please look carefully at ticket #33 and direct any > questions and/or critics to me. > > Thanks! > Torsten > > -- > bash$ :(){ :|:&};: --------------------- http://www.unixer.de/ ----- > Torsten Hoefler, Ph.D. | Postdoctoral Fellow > Open Systems Lab | Indiana University > 150 S. Woodlawn Ave. | Bloomington, IN, 474045, USA > Lindley Hall Room 135 | +01 (812) 855-3608 > _______________________________________________ > mpi-22 mailing list > mpi-22_at_[hidden] > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-22 > --------------------------------------------------------------------- > Intel GmbH > Dornacher Strasse 1 > 85622 Feldkirchen/Muenchen Germany > Sitz der Gesellschaft: Feldkirchen bei Muenchen > Geschaeftsfuehrer: Douglas Lusk, Peter Gleissner, Hannes Schwaderer > Registergericht: Muenchen HRB 47456 Ust.-IdNr. > VAT Registration No.: DE129385895 > Citibank Frankfurt (BLZ 502 109 00) 600119052 > > This e-mail and any attachments may contain confidential material for > the sole use of the intended recipient(s). Any review or distribution > by others is strictly prohibited. If you are not the intended > recipient, please contact the sender and delete all copies. > > > _______________________________________________ > mpi-22 mailing list > mpi-22_at_[hidden] > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-22 William Gropp Deputy Director for Research Institute for Advanced Computing Applications and Technologies Paul and Cynthia Saylor Professor of Computer Science University of Illinois Urbana-Champaign

1 0

Re: [Mpi-22] Ticket #33 (Scalable Graph Topology Interface)
by unknown＠example.com 18 Jun '09

18 Jun '09

Hi, Thanks. I'm not sure what part of the Wiki is meant when the guidelines are mentioned. Is it "Some guidelines for submissions" at the top of https://svn.mpi-forum.org/trac/mpi-forum-web/wiki/MpiTwoTwoWikiPage ? If so, I'm can't see ABI compatibility there yet, only backward compatibility in the sense that "it must not break existing correct programs" that may be understood as "recompile and enjoy". However, a forced recompilation is not what some customers will enjoy. Something mode definite, like "The intention of the Forum is that MPI-2.2 C and Fortran interfaces stay binary compatible with their MPI-2.1 counterparts" would probably suffice to address this. Note that C++ is intentionally omitted here. Best regards. Alexander -----Original Message----- From: mpi-22-bounces_at_[hidden] [mailto:mpi-22-bounces_at_[hidden]] On Behalf Of Torsten Hoefler Sent: Thursday, June 18, 2009 8:18 PM To: MPI 2.2 Subject: Re: [Mpi-22] Ticket #33 (Scalable Graph Topology Interface) On Thu, Jun 18, 2009 at 12:52:48PM -0500, William Gropp wrote: > I've added my recollection of our various discussions on this to the > wiki. These are the ones that I've applied as chair of the MPI 2.2 > effort. Thank you very much for this clarification Bill! I think/hope we all agree on this. Ticket #33 fulfills guideline 3. I also attached a BSD-like (that's what the lawyers at IU said) license file to all my implementations (#24, #31, #33, #94). Could everybody who is concerned with licensing issues (Alexander, Erez) please check the license and tell me if this is ok for you? Everybody else, please look carefully at ticket #33 and direct any questions and/or critics to me. Thanks! Torsten -- bash$ :(){ :|:&};: --------------------- http://www.unixer.de/ ----- Torsten Hoefler, Ph.D. | Postdoctoral Fellow Open Systems Lab | Indiana University 150 S. Woodlawn Ave. | Bloomington, IN, 474045, USA Lindley Hall Room 135 | +01 (812) 855-3608 _______________________________________________ mpi-22 mailing list mpi-22_at_[hidden] http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-22 --------------------------------------------------------------------- Intel GmbH Dornacher Strasse 1 85622 Feldkirchen/Muenchen Germany Sitz der Gesellschaft: Feldkirchen bei Muenchen Geschaeftsfuehrer: Douglas Lusk, Peter Gleissner, Hannes Schwaderer Registergericht: Muenchen HRB 47456 Ust.-IdNr. VAT Registration No.: DE129385895 Citibank Frankfurt (BLZ 502 109 00) 600119052 This e-mail and any attachments may contain confidential material for the sole use of the intended recipient(s). Any review or distribution by others is strictly prohibited. If you are not the intended recipient, please contact the sender and delete all copies.

1 0

Re: [Mpi-22] Ticket #33 (Scalable Graph Topology Interface)
by unknown＠example.com 18 Jun '09

18 Jun '09

On Thu, Jun 18, 2009 at 12:52:48PM -0500, William Gropp wrote: > I've added my recollection of our various discussions on this to the > wiki. These are the ones that I've applied as chair of the MPI 2.2 > effort. Thank you very much for this clarification Bill! I think/hope we all agree on this. Ticket #33 fulfills guideline 3. I also attached a BSD-like (that's what the lawyers at IU said) license file to all my implementations (#24, #31, #33, #94). Could everybody who is concerned with licensing issues (Alexander, Erez) please check the license and tell me if this is ok for you? Everybody else, please look carefully at ticket #33 and direct any questions and/or critics to me. Thanks! Torsten -- bash$ :(){ :|:&};: --------------------- http://www.unixer.de/ ----- Torsten Hoefler, Ph.D. | Postdoctoral Fellow Open Systems Lab | Indiana University 150 S. Woodlawn Ave. | Bloomington, IN, 474045, USA Lindley Hall Room 135 | +01 (812) 855-3608

1 0

Re: [Mpi-22] Ticket #33 (Scalable Graph Topology Interface)
by unknown＠example.com 18 Jun '09

18 Jun '09

I've added my recollection of our various discussions on this to the wiki. These are the ones that I've applied as chair of the MPI 2.2 effort. Bill On Jun 18, 2009, at 6:03 AM, Jeff Squyres wrote: > On Jun 18, 2009, at 3:47 AM, Supalov, Alexander wrote: > >> I wonder whether we may want to consider defining a prototype more >> strictly, say, by adding a couple of qualifiers like "demonstrating >> the intended functionality and performance potential" or something. >> This will definitely increase the quality of the prototypes, and may >> positively influence the quality of the resulting standard, which is >> after all what all of us are after here. Given that most of the >> intended MPI-2.2 tickets already appear to have good prototypes, it >> should not be a problem to clarify the rules a bit here. >> > > > Actually, it would be better to make these "ad hoc" rules more > definitive. Torsten is the latest in the series of people who have > been bitten by thinking that they were acting by the rules but then > someone said, "no, the rule is actually ..." > > Having somewhat subjective rules is fine, but having nebulous, not- > written-down, sometimes-seemingly-arbitrary rules is what is causing a > problem. > > How about someone making a shot at writing down exactly what the rules > are? Leaving leeway for some subjectivity is fine (and good), but at > leaving having *something* to point to when issues come under debate > like this would be most helpful -- and will save a lot of time during > the MPI-3 process. > > -- > Jeff Squyres > Cisco Systems > > _______________________________________________ > mpi-22 mailing list > mpi-22_at_[hidden] > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-22 William Gropp Deputy Director for Research Institute for Advanced Computing Applications and Technologies Paul and Cynthia Saylor Professor of Computer Science University of Illinois Urbana-Champaign

1 0