February 2013 - mpi-forum - mailman3.cels.anl.gov

Re: [Mpi-forum] MPI Forum voting rules: new version
by Jeff Squyres (jsquyres) 11 Feb '13

11 Feb '13

On Jan 17, 2013, at 6:02 AM, Rolf Rabenseifner <rabenseifner(a)hlrs.de> wrote: > Here is a list of small change requests/proposals: > > Page 3, one can read: > Here is the description of the process for > updating the MPI standard for MPI-3. > Please change "MPI-3" > into "the next version of the MPI standard" That is the Bill / Rich / Torsten document. I don't plan to update that chapter. It was specific to MPI-3. > Page 4, one can read > This formal reading requires also the make reviewchangeonlydoc pdf > file, which has identical text content as the final document, but all > changes since MPI-2.2 are still colored. > Please change "since MPI-2.2" > into "since the current official version of MPI". > > Page 5, one can read: > SVN Revision 691 is identical to the MPI-3.0 Draft > document from Nov. 12, 2010. > but should read > SVN Revision 1627 is identical to the final MPI-3.0 document > from Sep. 21, 2012 (it includes late changes that are part > of the official document as it is on the web (and printed)). > Revision 1627 still includes all change-macros. > The basis for next MPI will be a cleaned SVN revision, i.e., > without all these change-macros. > This cleaned version does not yet exist. Ditto for all of the above -- I don't plan to change that part of the document. Someone else is free to do so, but it's in the past, so I'm not sure if it's a good use of time to do so. To be clear: the Forum requested that Brian and I include that document in our proposal so that people could see/compare. I do not consider that text to be part of our proposal. > Page 8, 5. (c) reads: > (c) The organization had at least one of its > representatives physically present at any time during > that specific physical MPI Forum meeting. > The wording "any time" seems to be not intended. > It may be substituted by: > (c) The organization started to have at least one of its > representatives physically present at > that specific physical MPI Forum meeting. Sure. > Rolf > > ----- Original Message ----- >> From: "Jeff Squyres (jsquyres)" <jsquyres(a)cisco.com> >> To: "MPI Forum list" <mpi-forum(a)lists.mpi-forum.org> >> Sent: Wednesday, January 16, 2013 9:24:18 PM >> Subject: [Mpi-forum] MPI Forum voting rules: new version >> We had a very good discussion about the proposed MPI Forum voting >> rules at the meeting in December. >> >> Attached is a new version of the document from me/Brian/Martin with >> all the feedback from the December meeting: changes are listed in red. >> >> The biggest change is that we added a chapter 3: Suggestions for >> Voters. It's a 1-page bullet list of things that people voting should >> take into account when voting on proposals. We also moved the concept >> of time limits on proposals to this chapter (so time limits are now no >> longer binding -- they're just a suggestion). >> >> Please let us know your feedback on this document. >> >> Martin: can we get some time to discuss / vote on this document next >> meeting? >> >> -- >> Jeff Squyres >> jsquyres(a)cisco.com >> For corporate legal information go to: >> http://www.cisco.com/web/about/doing_business/legal/cri/ >> >> _______________________________________________ >> 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: Room 1.307) > _______________________________________________ > mpi-forum mailing list > mpi-forum(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum -- Jeff Squyres jsquyres(a)cisco.com For corporate legal information go to: http://www.cisco.com/web/about/doing_business/legal/cri/

1 0

Hotel rooms for the March Meeting
by Richard Graham 11 Feb '13

11 Feb '13

Just a reminder that Friday, Feb 15th, is the last day to book a room at the Renaissance hotel in downtown Chicago. Rich

1 0

Re: [Mpi-forum] MPI Forum voting rules: new version
by Jeff Squyres (jsquyres) 11 Feb '13

11 Feb '13

On Jan 16, 2013, at 4:36 PM, Aurï¿½lien Bouteiller <bouteill(a)icl.utk.edu> wrote: > Maybe the individual ballot quorum should be defined with respect to the number of abstain only, and not as a function of IMOVE institutions that may leave without setting voting instructions. What about redefining rule 7. as "abstain votes must not represent more than 1/4 of total votes on that particular ballot" ? Abstaining is the act of not voting (i.e., it's the same as not saying anything when the secretary calls your organization to vote). So I'm not quite sure what you mean -- those who leave early are implicitly abstaining (see para 3 in 2.3.1). I don't see the difference between what you propose and what is in rule 7. > I also still think that the 7. rule of less than 1/4 abstain is too high. the 3/4 of yes/no is already ensuring strong consensus among people who care. The point of rule 7 is to ensure that enough people care. > As is, that 7 rule is just a convoluted rewrite of the rule that abstain/no are the same thing Abstain != no in this proposal. See 2.3.1, bullets #1 and #2. > (to pass, a ballot has to get 1/2 of the IMOVE quorum). This is especially problematic if the definition of IMOVE includes a lot of people who cannot cast votes because they are not present anymore physically. Simple solution: ensure that you *are* present to vote. :-) To be clear: I do not intend to alter this proposal to cater to organizations who do not plan to be present to vote. > But even in general terms, it may be interesting to have "abstain" be a vote that has less impact than "no" on the success of a ballot, a flexibility given by setting the bar of abstain to less than 1/3 or 1/2 (or any fraction in-between). I repeat what I said above: abstain != no in this proposal. I think I have been pretty public in my feeling that abstain==no is a terrible idea. :-) -- Jeff Squyres jsquyres(a)cisco.com For corporate legal information go to: http://www.cisco.com/web/about/doing_business/legal/cri/

1 0