Folks, I am worried we will be pidgeonholed into a specific list of concerns and not take the big picture view, and then organize along the lines of solving several levels of problems.
There ought to be a high-level vision of what kind of activities strategic as well as tactical activities this standards group will do, rather than just a list of problems with MPI-2.1 at the implementor's level.
Here is a suggested set
1) Work to make corrections to the existing MPI-1.2, and MPI-2.1 specifications
2) Minor corrections and additions of a limited scope to address key deficiencies noted in specific areas, keyed to making an MPI-1.3, or MPI-2.2 level of change... goal of completion : 1 year.
* Strategic design study towards an MPI-2010
We owe it to ourselves and the intended audience(s) to answer questions including:
* Should this group even reconvene ? :-)
* Who will be positively impacted if a new version of MPI is made or not made? Presuming 'yes', where is the most 'bang for the buck'?
* Who is the audience for each activity (labs, industry, worldwide user base of existing MPI standards, etc)
* What existing and new constituencies do we plan to serve (multicomputer/HPC, Grid?, massively parallel cores, streaming and heterogeneous architectures, etc)
* How do we avoid the second system syndrome if we make another major release of MPI
* How do we support profiles that are relevant to application spaces? For instance : Is there any interest in embedded, realtime, fault resistant or DOD-type domains this time around?
* What is the right organization and participants for each kind of activity? How do we ensure inclusiveness and yet make progress?
* Do we want official standards sanction this time around? (IEEE, ISO, etc)
* Is there something smarter than an object-based API and LIS that can be defined, given the passage of time and the ascendancy of U.P., object modeling, etc?
With kind regards,
Tony
----- Original Message -----
From: "Kannan Narasimhan" <kannan.narasimhan@xxxxxx>
To: mpi-21@xxxxxxxxxxxxx
Sent: Friday, October 5, 2007 12:34:38 PM (GMT-0600) America/Chicago
Subject: RE: Meeting at SC'07
Rich,
I'll be attending SC07 (representing HP), and very much appreciate this initiative to restart of MPI forum activities at this conference.
As I put together the list of topics of interest to send over to you, I'd like to check first if we plan to allocate a chunk of this meeting for a technical MPI 2.1 Errata discussion. There are some topics of extreme interest to us (like support for messages larger than 2GB) and I'm interested in getting the pulse of the forum members on such topics.
Thanx!
Kannan
From: owner-mpi-21@xxxxxxxxxxxxx [mailto:owner-mpi-21@xxxxxxxxxxxxx] On Behalf Of Richard Graham
Sent: Friday, October 05, 2007 2:55 AM
To: mpi-21@xxxxxxxxxxxxx
Subject: Re: Meeting at SC'07
I will have an agenda out in a week or so, so if people have specific items they want
to include, please send them to the list. The intent of this meeting is to get organized
for actively restarting MPI-forum activities, so items along those lines, and maybe a
list of initial topics of interest (not for technical discussion on 11/14) would be good
topics.
Thanks,
Rich
On 10/5/07 12:30 AM, "Erez Haba" <erezh@xxxxxxxxxxxxx> wrote:
Thanks Rich,
Please do let us know in advance. I did not plan to attend SC'07, but I would come for the day only for this meeting.
Thanks,
.Erez
-----Original Message-----
From: owner-mpi-21@xxxxxxxxxxxxx [mailto:owner-mpi-21@xxxxxxxxxxxxx] On Behalf Of Anthony Skjellum
Sent: Thursday, October 04, 2007 10:57 AM
To: mpi-21@xxxxxxxxxxxxx
Subject: Re: Meeting at SC'07
Rich, It would be good to know in advance
* agenda/goals
* likely attendees/who they represent
Thanks,
Tony
On Oct 4, 2007, at 9:56 AM, Richard Graham wrote:
> At the Euro PVM/MPI meeting in Paris, the participants agreed to
> hold a
> meeting at SC'07 to help restart MPI-Forum activities. I am making
> arrangements for this meeting, and am targeting to have this on
> Wed, Nov.
> 14th 5:30pm - 7:00 pm in Reno. I know that this will not be a good
> time for
> all, but looking over the list of BOF's this seems to be the best
> time to do
> this. I will let people know when details are firmed up, but
> wanted to let
> people know about the current tentative time, and ask people to start
> passing the word around.
>
> Thanks,
> Rich
--
The information contained in this communication may be confidential and is
intended only for the use of the recipient(s) named above. If the reader
of this communication is not the intended recipient(s), you are hereby
notified that any dissemination, distribution, or copying of this
communication, or any of its contents, is strictly prohibited. If you are
not a named recipient or received this communication by mistake, please
notify the sender and delete the communication and all copies of it.