[Mpi-forum] [EXTERNAL] Re: Voting in July (and beyond)

Rusty Lusk lusk at mcs.anl.gov
Fri Jun 15 13:06:27 CDT 2012

This note is to say that yes, the Steering Committee is alive.

I have not been attending MPI-3 meetings because of other responsibilities and because the Argonne group is well represented without me.  Nonetheless I am a keen follower of Forum activities, including the discussion in this thread.  So far I don't believe anyone has said anything crazy;  there seems to be a consensus that a simple majority of a very small number of people is insufficient for making important decisions.

In MPI-1 and -2 we had a different structure (chapters vs. tickets) which seemed to lead to involvement of a greater percentage of attendees studying each issue in the chapter working groups, and therefore fewer serious (non-straw) votes in the long run.  This made the "bad" situation very rare, and led to a certain acceptable level of informality about the voting rules.  Nonetheless, it did happen from time to time that the "bad" situation did arise, and we would have an unusually large number of abstentions.  When this happened we would say to ourselves, "Well, we thought this was ready for a vote but apparently it isn't" and the issue would be deferred until more people felt comfortable taking a stand either for or against.  The "Journal of Development" ended up containing many interesting ideas that essentially got deferred off the end of the process, and most people thought that this was an appropriate result for proposals that did not generate sufficient enthusiasm either for their inclusion or exclusion.

The very nature of MPI-3 is more difficult than MPI-1 or -2, due to the interlocking of issues like fault tolerance, etc., so I don't think there is anything wrong with the current structure for topics.  The challenge is to find a way to define a process that achieves what everyone wants (robust support for MPI-3) without getting hamstrung in rule-making.  

(Hey, grizzled veterans, I'll bet there is a paper or two in examining the venerable JOD from the archives.  "What would have been the consequences if this or that proposal had become part of the MPI standard?")


On Friday,Jun 15, 2012, at 10:34 AM, Jeff Squyres wrote:

> On Jun 15, 2012, at 9:47 AM, Richard Graham wrote:
>> While some may view this as a change in rules, this was in fact a clarification.
> Fair enough: the rules themselves did not change.  What happened in Japan was a different interpretation than what has been used in the past.
>> This seems like a good topic to bring up to the Steering Committee.
> The meetings page lists the steering committee as:
> • Jack Dongarra
> • Al Geist
> • Richard Graham
> • Bill Gropp
> • Andrew Lumsdaine
> • Ewing Lusk
> • Rolf Rabenseifner
> At one point, you told me that, as the secretary, I was part of the Steering Committee, too.  That was not documented, however.
> FWIW: Of the list above list, 4 of the 7 are not active in the Forum.
> -- 
> Jeff Squyres
> jsquyres at cisco.com
> For corporate legal information go to: http://www.cisco.com/web/about/doing_business/legal/cri/
> _______________________________________________
> mpi-forum mailing list
> mpi-forum at lists.mpi-forum.org
> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum

More information about the mpi-forum mailing list