December 2010 - mpi-forum - mailman3.cels.anl.gov

Minor Changes to Profiling Chapter
by Bronis R. de Supinski 26 Dec '10

26 Dec '10

All: In preparation for the proposals coming out of the Tools Working Group, the Profiling Chapter Committee (i.e., Martin, Kathryn and me) are proposing several small grammatical changes to the existing text. We will cover these changes at the February meeting with the intent to obtain a vote to update the golden copy. The attached file shows all of the proposed changes. The attached file is also available off the Tools WG page or directly from: https://svn.mpi-forum.org/trac/mpi-forum-web/raw-attachment/wiki/MPI3Tools/… Please let me know if you have comments or concerns. Bronis

1 0

Re: [Mpi-forum] Proposal for the MPI-3 standardization process
by Supalov, Alexander 24 Dec '10

24 Dec '10

... and probably clarify in the process document the acceptable forms of implementation, delivery form, and licenses, if applicable. -----Original Message----- From: mpi-forum-bounces(a)lists.mpi-forum.org [mailto:[email protected]] On Behalf Of Underwood, Keith D Sent: Wednesday, December 22, 2010 6:47 PM To: Main MPI Forum mailing list Subject: Re: [Mpi-forum] Proposal for the MPI-3 standardization process I must say that I think that keeping the implementation requirement strong is critical at this phase. The current standard is complex and implementation issues are likely to be subtle. I don't think we have been through a single meeting where any one person has been "right" about their remembrance of the details of the standard. If an implementation is such a burden that it is an impediment to getting something in, that raises a lot of questions about whether it should get in. After all, we presumably want implementations to include these features in the foreseeable future, right? That said, I think "unanimous" is probably slight overkill. It would allow a single individual a procedural maneuver to kill (or substantially delay) minor changes to the standard or to a proposal. I also think we will have to look at what qualifies as an "implementation" on a case by case basis. Keith ________________________________________ From: mpi-forum-bounces(a)lists.mpi-forum.org [mpi-forum-bounces(a)lists.mpi-forum.org] On Behalf Of Graham, Richard L. [rlgraham(a)ornl.gov] Sent: Wednesday, December 22, 2010 10:32 AM To: Main MPI Forum mailing list Subject: Re: [Mpi-forum] Proposal for the MPI-3 standardization process I will take this, and what Rolf sent, and add to what we had decided at the meeting. I think the main point of disagreement is on the implementation. There was a very strong sense the an implementation is required before formal action by the forum. However, since this is intended to help the process, and not just create work, the notion was that the forum could decide to bypass this, but it would have to be a unanimous decision. Rich On 12/22/10 12:04 PM, "William Gropp" <wgropp(a)illinois.edu> wrote: I have taken Rolf's version and further edited it to clarify the language a bit and add the definition of simple majority that we have always used. Bill _______________________________________________ mpi-forum mailing list mpi-forum(a)lists.mpi-forum.org http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum _______________________________________________ mpi-forum mailing list mpi-forum(a)lists.mpi-forum.org http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum -------------------------------------------------------------------------------------- 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: [Mpi-forum] Proposal for the MPI-3 standardization process
by William Gropp 22 Dec '10

22 Dec '10

I'll reiterate that requiring an implementation is a major change over the MPI 1 and MPI 2 process. In addition, I'm not sure what real value there is in requiring an implementation that runs on a particular platform. In the 2.x process, this was intended to address the "minimal implementation effort" requirement, and hence was a reasonable requirement. This should not be a constraint for MPI-3. Bill On Dec 22, 2010, at 11:32 AM, Graham, Richard L. wrote: > I will take this, and what Rolf sent, and add to what we had decided > at the meeting. I think the main point of disagreement is on the > implementation. There was a very strong sense the an implementation > is required before formal action by the forum. However, since this > is intended to help the process, and not just create work, the > notion was that the forum could decide to bypass this, but it would > have to be a unanimous decision. > > Rich > > > On 12/22/10 12:04 PM, "William Gropp" <wgropp(a)illinois.edu> wrote: > > I have taken Rolf's version and further edited it to clarify the > language a bit and add the definition of simple majority that we have > always used. > > Bill > > > > _______________________________________________ > mpi-forum mailing list > mpi-forum(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum 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-forum] Proposal for the MPI-3 standardization process
by Underwood, Keith D 22 Dec '10

22 Dec '10

I must say that I think that keeping the implementation requirement strong is critical at this phase. The current standard is complex and implementation issues are likely to be subtle. I don't think we have been through a single meeting where any one person has been "right" about their remembrance of the details of the standard. If an implementation is such a burden that it is an impediment to getting something in, that raises a lot of questions about whether it should get in. After all, we presumably want implementations to include these features in the foreseeable future, right? That said, I think "unanimous" is probably slight overkill. It would allow a single individual a procedural maneuver to kill (or substantially delay) minor changes to the standard or to a proposal. I also think we will have to look at what qualifies as an "implementation" on a case by case basis. Keith ________________________________________ From: mpi-forum-bounces(a)lists.mpi-forum.org [mpi-forum-bounces(a)lists.mpi-forum.org] On Behalf Of Graham, Richard L. [rlgraham(a)ornl.gov] Sent: Wednesday, December 22, 2010 10:32 AM To: Main MPI Forum mailing list Subject: Re: [Mpi-forum] Proposal for the MPI-3 standardization process I will take this, and what Rolf sent, and add to what we had decided at the meeting. I think the main point of disagreement is on the implementation. There was a very strong sense the an implementation is required before formal action by the forum. However, since this is intended to help the process, and not just create work, the notion was that the forum could decide to bypass this, but it would have to be a unanimous decision. Rich On 12/22/10 12:04 PM, "William Gropp" <wgropp(a)illinois.edu> wrote: I have taken Rolf's version and further edited it to clarify the language a bit and add the definition of simple majority that we have always used. Bill _______________________________________________ mpi-forum mailing list mpi-forum(a)lists.mpi-forum.org http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum

1 0

Re: [Mpi-forum] Proposal for the MPI-3 standardization process
by Graham, Richard L. 22 Dec '10

22 Dec '10

I will take this, and what Rolf sent, and add to what we had decided at the meeting. I think the main point of disagreement is on the implementation. There was a very strong sense the an implementation is required before formal action by the forum. However, since this is intended to help the process, and not just create work, the notion was that the forum could decide to bypass this, but it would have to be a unanimous decision. Rich On 12/22/10 12:04 PM, "William Gropp" <wgropp(a)illinois.edu> wrote: I have taken Rolf's version and further edited it to clarify the language a bit and add the definition of simple majority that we have always used. Bill

1 0

Re: [Mpi-forum] Proposal for the MPI-3 standardization process
by William Gropp 22 Dec '10

22 Dec '10

I have taken Rolf's version and further edited it to clarify the language a bit and add the definition of simple majority that we have always used. Bill On Dec 11, 2010, at 6:39 AM, Rolf Rabenseifner wrote: > I had to leave earlier, therefore I do not know the result > of the discussion on Wednesday. > > I attach a version that should reflect Bill's comment and > that includes my additional ideas about the role of > - the chapter votes, > - the SVN trunk, and > - the reporting rules based on the ticket system. > > Rolf > > ----- Original Message ----- >> From: "William Gropp" <wgropp(a)illinois.edu> >> To: "Main MPI Forum mailing list" <mpi-forum(a)lists.mpi-forum.org> >> Sent: Wednesday, December 8, 2010 6:09:05 PM >> Subject: Re: [Mpi-forum] Proposal for the MPI-3 standardization >> process >> The requirement for an implementation is understandable but is a >> departure from MPI 1 and MPI 2 - there we only required a convincing >> argument that the requirements were possible, not an example code. I >> think this worked reasonably well (with some pluses and minus) and >> believe that the MPI 1 and MPI 2 rule should be sufficient. >> >> Bill > > > ----- Original Message ----- >> From: "Richard L. Graham" <rlgraham(a)ornl.gov> >> To: "Main MPI Forum mailing list" <mpi-forum(a)lists.mpi-forum.org> >> Sent: Wednesday, December 8, 2010 2:29:19 AM >> Subject: [Mpi-forum] Proposal for the MPI-3 standardization process >> The attached document is a summary of our discussion yesterday - we >> will talk about this tomorrow (Wed) as see if there are updates that >> should be made. Please come prepared with comments. >> >> Thanks, >> Rich >> >> _______________________________________________ >> mpi-forum mailing list >> mpi-forum(a)lists.mpi-forum.org >> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum > > -- > 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) > <standardization_process.docx><ATT00001..txt> 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-forum] Proposal for the MPI-3 standardization process
by Rolf Rabenseifner 11 Dec '10

11 Dec '10

I had to leave earlier, therefore I do not know the result of the discussion on Wednesday. I attach a version that should reflect Bill's comment and that includes my additional ideas about the role of - the chapter votes, - the SVN trunk, and - the reporting rules based on the ticket system. Rolf ----- Original Message ----- > From: "William Gropp" <wgropp(a)illinois.edu> > To: "Main MPI Forum mailing list" <mpi-forum(a)lists.mpi-forum.org> > Sent: Wednesday, December 8, 2010 6:09:05 PM > Subject: Re: [Mpi-forum] Proposal for the MPI-3 standardization process > The requirement for an implementation is understandable but is a > departure from MPI 1 and MPI 2 - there we only required a convincing > argument that the requirements were possible, not an example code. I > think this worked reasonably well (with some pluses and minus) and > believe that the MPI 1 and MPI 2 rule should be sufficient. > > Bill ----- Original Message ----- > From: "Richard L. Graham" <rlgraham(a)ornl.gov> > To: "Main MPI Forum mailing list" <mpi-forum(a)lists.mpi-forum.org> > Sent: Wednesday, December 8, 2010 2:29:19 AM > Subject: [Mpi-forum] Proposal for the MPI-3 standardization process > The attached document is a summary of our discussion yesterday - we > will talk about this tomorrow (Wed) as see if there are updates that > should be made. Please come prepared with comments. > > Thanks, > Rich > > _______________________________________________ > mpi-forum mailing list > mpi-forum(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum -- 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

Please send me your slides
by Jeff Squyres 09 Dec '10

09 Dec '10

>From the San Jose meeting this week. I'll post them on the web site. Thanks! -- Jeff Squyres jsquyres(a)cisco.com For corporate legal information go to: http://www.cisco.com/web/about/doing_business/legal/cri/

1 0

Re: [Mpi-forum] Proposal for the MPI-3 standardization process
by William Gropp 08 Dec '10

08 Dec '10

The requirement for an implementation is understandable but is a departure from MPI 1 and MPI 2 - there we only required a convincing argument that the requirements were possible, not an example code. I think this worked reasonably well (with some pluses and minus) and believe that the MPI 1 and MPI 2 rule should be sufficient. Bill On Dec 7, 2010, at 7:29 PM, Graham, Richard L. wrote: > The attached document is a summary of our discussion yesterday - we > will talk about this tomorrow (Wed) as see if there are updates that > should be made. Please come prepared with comments. > > Thanks, > Rich > <standardization_process.docx><ATT00001..txt> 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

Proposal for the MPI-3 standardization process
by Graham, Richard L. 07 Dec '10

07 Dec '10

The attached document is a summary of our discussion yesterday - we will talk about this tomorrow (Wed) as see if there are updates that should be made. Please come prepared with comments. Thanks, Rich

1 0