May 2011 - Mpi3-bwcompat - mailman3.cels.anl.gov

Notes from last meeting
by unknown＠example.com 25 May '11

25 May '11

We met on 5/20 to discuss the future of the WG. We decided to continue to meet and will resume our normal biweekly schedule starting 6/3. The following items were areas we may wish to contribute. During our next meeting we will discuss these further, particularly those areas being actively worked by members of the WG: Removing C++ or make C++ optional Removing Deprecated functions MPI_Count Datatype creation functions MPI_Count for pt2pt/collectives? Timer requests Non-blocking connect/accept and/or connect/accept with timeout Defining mpirun return value mpirun -quiet Thanks, Dave

1 0

Re: [Mpi3-bwcompat] Phone conf schedule?
by unknown＠example.com 18 May '11

18 May '11

I didn't send the MPI_Timer stuff to Rich because I don't think it needs to be a gating feature of MPI-3.0. On May 17, 2011, at 2:29 PM, Fab Tillier wrote: > Solt, David George wrote on Tue, 17 May 2011 at 11:24:37 > >> One other cleanup point. Rich requested: >> >>> One more thing. Those of you that have an item on the list (and those >>> that are adding items) - please send me a 1-2 sentence explanation that >>> we can use as a >> description." >> >> Has anyone sent him anything yet? > > I sent him my 3 new items. I don't know if Jeff sent him the timer stuff or not. > >> I am open to working on other miscellaneous or backward compatibility >> related issues. I could schedule a one-time meeting to discuss our future >> and then decide if we should continue with a re-occurring meeting? > > Sure, that sounds good. Same time as usual, this Friday? > > -Fab > >> >> Thanks, >> Dave >> >> -----Original Message----- >> From: mpi3-bwcompat-bounces_at_[hidden] [mailto:mpi3- >> bwcompat-bounces_at_[hidden]] On Behalf Of Fab Tillier >> Sent: Tuesday, May 17, 2011 1:17 PM >> To: MPI-3 backwards compatability WG >> Subject: Re: [Mpi3-bwcompat] Phone conf schedule? >> >> Quincey Koziol wrote on Tue, 17 May 2011 at 11:09:15 >> >>> On May 17, 2011, at 1:01 PM, Fab Tillier wrote: >>> >>>> The meeting request I had expired 4/15/2011. Do we want to resume it? >>>> >>>> Couple topics of interest, actually on topic for our working group: - >>>> Removing deprecated functions (e.g. MPI_UB, MPI_LB, etc.) from MPI >>>> 3.0. They are still available via MPI 2.2 support, and an >>>> implementation can easily provide concurrent support for MPI 2.2 and >>>> MPI 3.0, should that be required. - Defining the C++ bindings as >>>> optional in MPI 3.0. This allows implementations that do not ship C++ >>>> bindings to be standard compliant. - Removing C++ bindings from MPI >>>> 3.0. This one goes counter to the previous one, but follows the same >>>> logic as for the other deprecated functions. If we don't provide C++ >>>> bindings for new MPI 3.0 functionality, we should remove them from MPI >>>> 3.0. They can still be supported via MPI 2.2 compliance. >>> >>> Anything to discuss about const buffers? >> >> We passed the first reading, so we're going for 1st vote at the next meeting. >> There was one minor ticket0 wording change for the IN description, but >> nothing major. I don't think there's much to discuss at this point. >> >>> I've got someone working on extensions to the datatype construction >>> routines, creating a use case/justification for why we should add those >>> routines to 3.0. Adam Moody was going to create a ticket for those, >>> once I give him this use case document. [Dunno if we need to talk >>> about anything yet on this though] >> >> That would be nice to see. Jeff and I are also working on the MPI Timer >> request stuff, don't know if we want to discuss that in this working group, I'd >> be up for it, I like this crowd... >> >> -Fab >> >> _______________________________________________ Mpi3-bwcompat mailing >> list Mpi3-bwcompat_at_[hidden] >> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-bwcompat >> >> _______________________________________________ >> Mpi3-bwcompat mailing list >> Mpi3-bwcompat_at_[hidden] >> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-bwcompat > > _______________________________________________ > Mpi3-bwcompat mailing list > Mpi3-bwcompat_at_[hidden] > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-bwcompat -- Jeff Squyres jsquyres_at_[hidden] For corporate legal information go to: http://www.cisco.com/web/about/doing_business/legal/cri/

1 0

Declined: MPI Backward Compatibility Working Group
by unknown＠example.com 18 May '11

18 May '11

-------------------------------------------------------------------------------------- Intel GmbH Dornacher Strasse 1 85622 Feldkirchen/Muenchen, Deutschland 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 a.M. (BLZ 502 109 00) 600119052 *

1 0

Declined: MPI Backward Compatibility Working Group
by unknown＠example.com 18 May '11

18 May '11

-------------------------------------------------------------------------------------- Intel GmbH Dornacher Strasse 1 85622 Feldkirchen/Muenchen, Deutschland 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 a.M. (BLZ 502 109 00) 600119052 *

1 0

Re: [Mpi3-bwcompat] Phone conf schedule?
by unknown＠example.com 17 May '11

17 May '11

Solt, David George wrote on Tue, 17 May 2011 at 11:33:12 > I'll setup the meeting for the usual time this Friday. > > Did anyone send a 1-2 sentence explanation of MPI_Count to Rich? Just sent them to Rich. -Fab > > Thanks, > Dave > > -----Original Message----- > From: mpi3-bwcompat-bounces_at_[hidden] [mailto:mpi3- > bwcompat-bounces_at_[hidden]] On Behalf Of Fab Tillier > Sent: Tuesday, May 17, 2011 1:29 PM > To: MPI-3 backwards compatability WG > Subject: Re: [Mpi3-bwcompat] Phone conf schedule? > > Solt, David George wrote on Tue, 17 May 2011 at 11:24:37 > >> One other cleanup point. Rich requested: >> >>> One more thing. Those of you that have an item on the list (and those >>> that are adding items) - please send me a 1-2 sentence explanation that >>> we can use as a >> description." >> >> Has anyone sent him anything yet? > > I sent him my 3 new items. I don't know if Jeff sent him the timer > stuff or not. > >> I am open to working on other miscellaneous or backward compatibility >> related issues. I could schedule a one-time meeting to discuss our future >> and then decide if we should continue with a re-occurring meeting? > > Sure, that sounds good. Same time as usual, this Friday? > > -Fab > >> >> Thanks, >> Dave >> >> -----Original Message----- >> From: mpi3-bwcompat-bounces_at_[hidden] [mailto:mpi3- >> bwcompat-bounces_at_[hidden]] On Behalf Of Fab Tillier >> Sent: Tuesday, May 17, 2011 1:17 PM >> To: MPI-3 backwards compatability WG >> Subject: Re: [Mpi3-bwcompat] Phone conf schedule? >> >> Quincey Koziol wrote on Tue, 17 May 2011 at 11:09:15 >> >>> On May 17, 2011, at 1:01 PM, Fab Tillier wrote: >>> >>>> The meeting request I had expired 4/15/2011. Do we want to resume it? >>>> >>>> Couple topics of interest, actually on topic for our working group: - >>>> Removing deprecated functions (e.g. MPI_UB, MPI_LB, etc.) from MPI >>>> 3.0. They are still available via MPI 2.2 support, and an >>>> implementation can easily provide concurrent support for MPI 2.2 and >>>> MPI 3.0, should that be required. - Defining the C++ bindings as >>>> optional in MPI 3.0. This allows implementations that do not ship >>>> C++ bindings to be standard compliant. - Removing C++ bindings from >>>> MPI 3.0. This one goes counter to the previous one, but follows the >>>> same logic as for the other deprecated functions. If we don't >>>> provide C++ bindings for new MPI 3.0 functionality, we should remove >>>> them from MPI 3.0. They can still be supported via MPI 2.2 >>>> compliance. >>> >>> Anything to discuss about const buffers? >> >> We passed the first reading, so we're going for 1st vote at the next >> meeting. There was one minor ticket0 wording change for the IN >> description, but nothing major. I don't think there's much to discuss >> at this point. >> >>> I've got someone working on extensions to the datatype construction >>> routines, creating a use case/justification for why we should add >>> those routines to 3.0. Adam Moody was going to create a ticket for >>> those, once I give him this use case document. [Dunno if we need to >>> talk about anything yet on this though] >> >> That would be nice to see. Jeff and I are also working on the MPI >> Timer request stuff, don't know if we want to discuss that in this >> working group, I'd be up for it, I like this crowd... >> >> -Fab >> >> _______________________________________________ Mpi3- bwcompat mailing >> list Mpi3-bwcompat_at_[hidden] >> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-bwcompat >> >> _______________________________________________ >> Mpi3-bwcompat mailing list >> Mpi3-bwcompat_at_[hidden] >> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-bwcompat > > _______________________________________________ Mpi3-bwcompat mailing > list Mpi3-bwcompat_at_[hidden] > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-bwcompat > > _______________________________________________ > Mpi3-bwcompat mailing list > Mpi3-bwcompat_at_[hidden] > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-bwcompat

1 0

Re: [Mpi3-bwcompat] [Mpi-forum] MPI 3.0
by unknown＠example.com 17 May '11

17 May '11

Hi Rich, Here's the description for the two items we have on the list: Const: make C bindings const-correct, except where backward compatibility would be broken. MPI_Count: Minimal support for large datatypes. -Fab From: mpi-forum-bounces_at_[hidden] [mailto:mpi-forum-bounces_at_[hidden]] On Behalf Of RICHARD GRAHAM Sent: Wednesday, May 11, 2011 8:30 PM To: mpi-forum_at_[hidden] Subject: Re: [Mpi-forum] MPI 3.0 One more thing. Those of you that have an item on the list (and those that are adding items) - please send me a 1-2 sentence explanation that we can use as a description. Thanks, Rich ________________________________ From: rlgraham32_at_[hidden] To: mpi-forum_at_[hidden] Date: Thu, 12 May 2011 01:46:58 +0100 Subject: [Mpi-forum] MPI 3.0 At the MPI forum meeting this week we decided to accelerate the release of MPI-3.0, and have a follow on MPI-3.1 when the more complete set of features is available. Currently, nonblocking collectives and MProbe have already been voted in. We have created a list of features that may be ready for first reading at the next couple of meetings, which will be used to determine the gating factors for the MPI-3.0 release. Here is the list we generated: - RMA - FT stabilization - Fortran bindings - MPI-T - Neighborhood collectives - Const - MPI Count - Nonblocking collective I/O - Hybrid: Shared memory communicator - Hybrid: Threads - Hybrid: Endpoints If people want to add more to this list, please let me know by end of the day Friday, 5/13/2011. Once we have the complete list, each eligible organization should vote for it's top three features, with respect to MPI-3.0. An organization is eligible to vote, if someone from that organization has participated in person at one or more meetings in 2011. The vote will take place 5/14-20/2011, and the top three items will be what gate the release of MPI-3.0. To vote, a representative of the organization should e-mail BOTH Jeff Squyres and myself with your vote. Other items that are voted in before 3.0 is final, will also be included. For MPI-3.1, we will change our process a bit, but that is still under discussion, and will be the subject of a future e-mail discussion. The eligible organizations include: - ANL - Bull - Cicso - Cray - Fujistsu - HDF - HLRS - HP - IBM - Indiana U. - INRIA - Intel - LLNL - LANL - Microsoft - NCSA/UIUC - ORNL - OSU - Oracle - SNL - U Alabama Birmingham - U Tennessee - U. Tokyo Rich _______________________________________________ mpi-forum mailing list mpi-forum_at_[hidden]<mailto:mpi-forum_at_[hidden]> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum *

1 0

MPI Backward Compatibility Working Group
by unknown＠example.com 17 May '11

17 May '11

When: Friday, May 20, 2011 12:00 PM-1:00 PM (GMT-06:00) Central Time (US & Canada). Where: 866.409.2889 id 3582114059 Note: The GMT offset above does not reflect daylight saving time adjustments. *~*~*~*~*~*~*~*~*~* U.S. toll number 702.696.4520 U.S. toll-free dial-in number International â see attached pdf 866.409.2889 - Conference Code: 3582114059 *

1 0

MPI Backward Compatibility Working Group
by unknown＠example.com 17 May '11

17 May '11

When: Occurs every 2 weeks on Friday effective 7/23/2010 until 5/21/2011 from 12:00 PM to 1:00 PM (GMT-06:00) Central Time (US & Canada). Where: 866.409.2889 id 3582114059 Note: The GMT offset above does not reflect daylight saving time adjustments. *~*~*~*~*~*~*~*~*~* U.S. toll number 702.696.4520 U.S. toll-free dial-in number International â see attached pdf 866.409.2889 - Conference Code: 3582114059 *

1 0

Re: [Mpi3-bwcompat] Phone conf schedule?
by unknown＠example.com 17 May '11

17 May '11

I'll setup the meeting for the usual time this Friday. Did anyone send a 1-2 sentence explanation of MPI_Count to Rich? Thanks, Dave -----Original Message----- From: mpi3-bwcompat-bounces_at_[hidden] [mailto:mpi3-bwcompat-bounces_at_[hidden]] On Behalf Of Fab Tillier Sent: Tuesday, May 17, 2011 1:29 PM To: MPI-3 backwards compatability WG Subject: Re: [Mpi3-bwcompat] Phone conf schedule? Solt, David George wrote on Tue, 17 May 2011 at 11:24:37 > One other cleanup point. Rich requested: > >> One more thing. Those of you that have an item on the list (and those >> that are adding items) - please send me a 1-2 sentence explanation that >> we can use as a > description." > > Has anyone sent him anything yet? I sent him my 3 new items. I don't know if Jeff sent him the timer stuff or not. > I am open to working on other miscellaneous or backward compatibility > related issues. I could schedule a one-time meeting to discuss our future > and then decide if we should continue with a re-occurring meeting? Sure, that sounds good. Same time as usual, this Friday? -Fab > > Thanks, > Dave > > -----Original Message----- > From: mpi3-bwcompat-bounces_at_[hidden] [mailto:mpi3- > bwcompat-bounces_at_[hidden]] On Behalf Of Fab Tillier > Sent: Tuesday, May 17, 2011 1:17 PM > To: MPI-3 backwards compatability WG > Subject: Re: [Mpi3-bwcompat] Phone conf schedule? > > Quincey Koziol wrote on Tue, 17 May 2011 at 11:09:15 > >> On May 17, 2011, at 1:01 PM, Fab Tillier wrote: >> >>> The meeting request I had expired 4/15/2011. Do we want to resume it? >>> >>> Couple topics of interest, actually on topic for our working group: - >>> Removing deprecated functions (e.g. MPI_UB, MPI_LB, etc.) from MPI >>> 3.0. They are still available via MPI 2.2 support, and an >>> implementation can easily provide concurrent support for MPI 2.2 and >>> MPI 3.0, should that be required. - Defining the C++ bindings as >>> optional in MPI 3.0. This allows implementations that do not ship C++ >>> bindings to be standard compliant. - Removing C++ bindings from MPI >>> 3.0. This one goes counter to the previous one, but follows the same >>> logic as for the other deprecated functions. If we don't provide C++ >>> bindings for new MPI 3.0 functionality, we should remove them from MPI >>> 3.0. They can still be supported via MPI 2.2 compliance. >> >> Anything to discuss about const buffers? > > We passed the first reading, so we're going for 1st vote at the next meeting. > There was one minor ticket0 wording change for the IN description, but > nothing major. I don't think there's much to discuss at this point. > >> I've got someone working on extensions to the datatype construction >> routines, creating a use case/justification for why we should add those >> routines to 3.0. Adam Moody was going to create a ticket for those, >> once I give him this use case document. [Dunno if we need to talk >> about anything yet on this though] > > That would be nice to see. Jeff and I are also working on the MPI Timer > request stuff, don't know if we want to discuss that in this working group, I'd > be up for it, I like this crowd... > > -Fab > > _______________________________________________ Mpi3-bwcompat mailing > list Mpi3-bwcompat_at_[hidden] > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-bwcompat > > _______________________________________________ > Mpi3-bwcompat mailing list > Mpi3-bwcompat_at_[hidden] > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-bwcompat _______________________________________________ Mpi3-bwcompat mailing list Mpi3-bwcompat_at_[hidden] http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-bwcompat

1 0

Re: [Mpi3-bwcompat] Phone conf schedule?
by unknown＠example.com 17 May '11

17 May '11

Solt, David George wrote on Tue, 17 May 2011 at 11:24:37 > One other cleanup point. Rich requested: > >> One more thing. Those of you that have an item on the list (and those >> that are adding items) - please send me a 1-2 sentence explanation that >> we can use as a > description." > > Has anyone sent him anything yet? I sent him my 3 new items. I don't know if Jeff sent him the timer stuff or not. > I am open to working on other miscellaneous or backward compatibility > related issues. I could schedule a one-time meeting to discuss our future > and then decide if we should continue with a re-occurring meeting? Sure, that sounds good. Same time as usual, this Friday? -Fab > > Thanks, > Dave > > -----Original Message----- > From: mpi3-bwcompat-bounces_at_[hidden] [mailto:mpi3- > bwcompat-bounces_at_[hidden]] On Behalf Of Fab Tillier > Sent: Tuesday, May 17, 2011 1:17 PM > To: MPI-3 backwards compatability WG > Subject: Re: [Mpi3-bwcompat] Phone conf schedule? > > Quincey Koziol wrote on Tue, 17 May 2011 at 11:09:15 > >> On May 17, 2011, at 1:01 PM, Fab Tillier wrote: >> >>> The meeting request I had expired 4/15/2011. Do we want to resume it? >>> >>> Couple topics of interest, actually on topic for our working group: - >>> Removing deprecated functions (e.g. MPI_UB, MPI_LB, etc.) from MPI >>> 3.0. They are still available via MPI 2.2 support, and an >>> implementation can easily provide concurrent support for MPI 2.2 and >>> MPI 3.0, should that be required. - Defining the C++ bindings as >>> optional in MPI 3.0. This allows implementations that do not ship C++ >>> bindings to be standard compliant. - Removing C++ bindings from MPI >>> 3.0. This one goes counter to the previous one, but follows the same >>> logic as for the other deprecated functions. If we don't provide C++ >>> bindings for new MPI 3.0 functionality, we should remove them from MPI >>> 3.0. They can still be supported via MPI 2.2 compliance. >> >> Anything to discuss about const buffers? > > We passed the first reading, so we're going for 1st vote at the next meeting. > There was one minor ticket0 wording change for the IN description, but > nothing major. I don't think there's much to discuss at this point. > >> I've got someone working on extensions to the datatype construction >> routines, creating a use case/justification for why we should add those >> routines to 3.0. Adam Moody was going to create a ticket for those, >> once I give him this use case document. [Dunno if we need to talk >> about anything yet on this though] > > That would be nice to see. Jeff and I are also working on the MPI Timer > request stuff, don't know if we want to discuss that in this working group, I'd > be up for it, I like this crowd... > > -Fab > > _______________________________________________ Mpi3-bwcompat mailing > list Mpi3-bwcompat_at_[hidden] > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-bwcompat > > _______________________________________________ > Mpi3-bwcompat mailing list > Mpi3-bwcompat_at_[hidden] > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-bwcompat

1 0