On Oct 1, 2012, at 4:58 AM, Rolf Rabenseifner wrote:
> Therefore, I would like that Jeff is doing a email voting
> by institutions after we fixed the questions:
I will not do a formal vote on this via email. There's all kinds of issues involved with that (e.g., who gets to vote?).
I suggest a much simpler solution. Please visit the following public Google spreadsheet and fill in the meeting location/date preferences for your organization:
https://docs.google.com/spreadsheet/ccc?key=0AtJBC25MGsp3dDd3bmhuRllybEpqRn…
This is informal, but please fill it out on a per-organization basis.
***********************************************
*** Please do so by COB, Monday, 8 Oct, 2012***
***********************************************
--
Jeff Squyres
jsquyres(a)cisco.com
For corporate legal information go to: http://www.cisco.com/web/about/doing_business/legal/cri/
Unlike many other cases where communicating >2 GB can be done with a
datatype workaround, MPI_PACK/UNPACK seem particularly vulnerable to
the 32-bit integer representation issue.
I don't know whether or not this has been discussed before (and
apologize if my message is redundant). Is there a proposal for
MPI_(UN)PACK_COUNT yet?
Thanks,
Jeff
---------- Forwarded message ----------
From: Christoph Sprenger <csprenger(a)wetafx.co.nz>
Date: Wed, Oct 3, 2012 at 3:10 AM
Subject: [mpich-discuss] MPI_Pack max buffer size
To: mpich-discuss(a)mcs.anl.gov
Hi,
just wanted to double check if there is a way to Pack/Unpack bigger
char buffers than numeric_limits<int>::max() without tokenizing the
message into multiple chunks ?
seems like one of my messages is hitting these bounds unfortunately.
Cheers,
Christoph
_______________________________________________
mpich-discuss mailing list mpich-discuss(a)mcs.anl.gov
To manage subscription options or unsubscribe:
https://lists.mcs.anl.gov/mailman/listinfo/mpich-discuss
--
Jeff Hammond
Argonne Leadership Computing Facility
University of Chicago Computation Institute
jhammond(a)alcf.anl.gov / (630) 252-5381
http://www.linkedin.com/in/jeffhammondhttps://wiki.alcf.anl.gov/parts/index.php/User:Jhammond
I need to know the final schedule for the MPI Forum meeting in December so that I can reserve meeting rooms.
Right now, I have one big room from noon on Monday through noon on Thursday.
Are we actually going to meet all the way through Thursday? (i.e., is there enough material / working group time requested for all the way through Thursday)
Do we need a breakout room? If so, for what timeslots? (it's hard to reserve a breakout room for an entire day; it's easier to reserve individual breakout rooms for segmented chunks of time)
--
Jeff Squyres
jsquyres(a)cisco.com
For corporate legal information go to: http://www.cisco.com/web/about/doing_business/legal/cri/
> So far, what I see from the mailing list and the in-person
> conversations earlier this week, is that most folks are in favor of
> keeping the connection. We can have a mail vote on this, or can
> postpone to the Dec meeting - no hurry on this one.
I need to fix my major dates more than one year in advance.
Therefore, I'm very late with September 2013 and would like
to have it fixed as soon as possible.
Therefore, I would like that Jeff is doing a email voting
by institutions after we fixed the questions:
I see two questions that need a formal decision:
A. The location:
- Sep 2013 meeting in USA
(currently proposed for Sep. 9-11 in San Jose),
- Sep 2013 before or after the EuroMPI in Madrid, Spain.
The EuroMPI is Sep. 15-18, 2013.
If we choose the second option, then we have mainly
the two options:
B. The exact date:
- The days directly after the EuroMPI,
i.e., Sep. 19-21, 2013 (which would imply together with the
flights (one/two days in the two directions from US/Asia) a
non-pause travel of 2+3+3+1=9 days when omitting the tutorial day)
- The week before with a free Saturday and three full days,
i.e., Wed-Fri, Sep. 11-13, 2013 (the probably more healthy option)
Best regards
Rolf
----- Original Message -----
> From: "Richard Graham" <richardg(a)mellanox.com>
> To: "Main MPI Forum mailing list" <mpi-forum(a)lists.mpi-forum.org>, "Bronis R. de Supinski" <bronis(a)llnl.gov>
> Sent: Friday, September 28, 2012 5:14:56 PM
> Subject: Re: [Mpi-forum] Meeting schedule for 2013
> I will add to this - there is NO formal relationship between the
> meeting and the conference. Both are independent bodies, that make
> their own decisions. EuroMPI does attract many MPI implementers, and
> since it is the only conference that focuses on MPI, and there is
> concern about it's viability if the two are decoupled. As an MPI
> developer, this is of great concern to me if EuroMPI goes away. Others
> expressed similar concern, and since we did not have time to discuss
> this at the last forum meeting, this is coming up on the mailing list,
> as we are trying to finalize on the meeting dates for next year.
>
> So far, what I see from the mailing list and the in-person
> conversations earlier this week, is that most folks are in favor of
> keeping the connection. We can have a mail vote on this, or can
> postpone to the Dec meeting - no hurry on this one.
>
> Rich
>
> -----Original Message-----
> From: mpi-forum-bounces(a)lists.mpi-forum.org
> [mailto:[email protected]] On Behalf Of Aurélien
> Bouteiller
> Sent: Friday, September 28, 2012 10:12 AM
> To: Bronis R. de Supinski; Main MPI Forum mailing list
> Subject: Re: [Mpi-forum] Meeting schedule for 2013
>
>
> Le 27 sept. 2012 à 08:47, "Bronis R. de Supinski" <bronis(a)llnl.gov> a
> écrit :
>
> >
> > Rolf:
> >
> > Re:
> >>> even years: 3US+euroMPI
> >>> odd years: 2US,1Asia+EuroMPI.
> >>
> >> Yes, this was the last position at the Forum and also discussed at
> >> the EuroMPI business meeting.
> >
> > As I understand the situation, this option was not discussed at the
> > Forum. As stated previously, the matter was decided in Chicago and
> > reopening it without an announcement was inappropriate.
> >
> > In addition, what is discussed at the EuroMPI business meeting is
> > not
> > relevant. It is not their decision.
> >
> Bronis, the matter has not been decided, as we are obviously still
> discussing it. Nothing was decided either in Vienna as it felt like
> there was no consensus still (straw votes where very divided, and many
> options came close to one another, we should be driven by consensus
> right?). No need to feel so strong on "rules" again, we are just
> expressing wishes so far.
>
> It so happens that many people are going to attend EuroMPI anyway, so
> collocating is a clear way of cutting cost for many attendees;
> however, it could be interesting to make so that we don't have to
> spend a WE in a foreign country when attending both the conference and
> the Forum, as this is expensive and annoying.
>
> Also, it is to be noted that european institutions participation is
> declining recently. Locating meetings in Asia was taken as a step to
> address the same issue when Asian participation was low. We might
> consider this in the equation, or risk that the forum becomes an US
> only body, and not an international body anymore.
>
> So the question is not in terms of us against them (the conference),
> it is in terms of what is convenient for the Forum members (many of
> whom do attend the conference, many of whom are not US based
> institutions) and makes sense for an -international- standardization
> body.
>
> Regards,
> Aurelien
>
> >> The MPI Forum was re-started by the activities at several EuroMPI
> >> conferences, the last one at Paris 2007.
> >>
> >> Therefore, there is a strong relationship and resposibility and to
> >> have the MPI Forum meeting in US and the EuroMPI in Europe both in
> >> fall, would not reflect this responsibility.
> >>
> >> As far as I see, the colleagues in Madrid would like to host the
> >> MPI
> >> Forum meeting together with the EuroMPI and therefore, I expect -
> >> based on the previous meeting rules - we will have
> >>
> >> MPI-next Forum meeting Sep. 11-13, 2013, (instead of Sep. 9-11 in
> >> San
> >> Jose), and EuroMPI, Sep. 15-18, 2013, both in Madrid, Spain
> >>
> >> This may change in later years if the EuroMPI may go to Asia, but
> >> this is another discussion within the EuroMPI organizers.
> >
> > What I hear is that holding the MPI Forum meeting separately from
> > EuroMPI would damage EuroMPI. Well, forcing the Forum to co-locate
> > is
> > putting the cart before the horse. If EuroMPI cannot withstand a
> > separate meeting, as decided by the Forum, then EuroMPI needs to
> > adapt
> > -- not the other way around.
> >
> > As was previously suggested, EuroMPI is free to rebrand as
> > International MPI and to co-locate with Forum meetings.
> > If the conference demonstrated that flexibility, I would be more
> > willing to consider their desires as an MPI Forum voting member.
> >
> > Bronis
> >
> >
> >
> >
> >> Best regards
> >> Rolf
> >>
> >> ----- Original Message -----
> >>> From: "Aurélien Bouteiller" <bouteill(a)eecs.utk.edu>
> >>> To: "Main MPI Forum mailing list" <mpi-forum(a)lists.mpi-forum.org>
> >>> Sent: Thursday, September 27, 2012 2:41:08 AM
> >>> Subject: Re: [Mpi-forum] Meeting schedule for 2013 The 3-1 with 1
> >>> alternate between asia and europe collected the most vote. However
> >>> there was significant disappointment about not collocating a
> >>> meeting
> >>> with EuroMPI, and after the straw vote closed, another interesting
> >>> possibility was envisioned:
> >>> even years: 3US+euroMPI
> >>> odd years: 2US,1Asia+EuroMPI.
> >>> This last proposal has not been straw voted, so it is not possible
> >>> to know if it has support, but it seemed from the reception it had
> >>> that the idea is not unbearable for most US institutions.
> >>> Aurelien
> >>> Le 26 sept. 2012 à 15:17, Jeff Squyres <jsquyres(a)cisco.com> a
> >>> écrit :
> >>> > I don't remember there being a majority for the 2-2 option here
> >>> > in
> >>> > Vienna -- am I remembering last week incorrectly?
> >>> >
> >>> > I definitely can't do 2 international meetings next week.
> >>> >
> >>> >
> >>> >
> >>> >
> >>> > On Sep 26, 2012, at 8:02 PM, Richard Graham wrote:
> >>> >
> >>> >> This was discussed and there was another vote at the meeting in
> >>> >> Vienna. This note is just a follow up to the discussions at
> >>> >> that
> >>> >> meeting.
> >>> >>
> >>> >> Rich
> >>> >>
> >>> >> -----Original Message-----
> >>> >> From: mpi-forum-bounces(a)lists.mpi-forum.org
> >>> >> [mailto:[email protected]] On Behalf Of
> >>> >> Schulz, Martin
> >>> >> Sent: Wednesday, September 26, 2012 7:36 PM
> >>> >> To: Bronis R. de Supinski; Main MPI Forum mailing list
> >>> >> Subject: Re: [Mpi-forum] Meeting schedule for 2013
> >>> >>
> >>> >> Hi Rich,
> >>> >>
> >>> >> I have the same recollection as Bronis and Mohamad. I thought
> >>> >> there was a strong majority (at least at the Chicago meeting)
> >>> >> for
> >>> >> a 3-1 solution with alternating non-US locations (one year
> >>> >> Asia,
> >>> >> one year Europe, but not both in one year - as it also shows on
> >>> >> the current tentative meeting calendar for next year). I know
> >>> >> this is not fair to everyone, but the group (again, at least in
> >>> >> Chicago, which was pretty complete, though, since we were
> >>> >> wrapping up 3.0) felt that this would maximize active
> >>> >> participation.
> >>> >>
> >>> >> Martin
> >>> >>
> >>> >>
> >>> >> On Sep 26, 2012, at 9:39 AM, Bronis R. de Supinski wrote:
> >>> >>
> >>> >>>
> >>> >>> Rich:
> >>> >>>
> >>> >>> I would have to disagree. At best, you are splitting hairs to
> >>> >>> say this option was not voted on. In the "alternate"
> >>> >>> years, it is simply the 2-2 option that was voted on and did
> >>> >>> not
> >>> >>> come anywhere near the 3-1 option in terms of votes.
> >>> >>>
> >>> >>> The arguments against 2-2 (even in alternating years) still
> >>> >>> apply.
> >>> >>> It
> >>> >>> significantly lowers attendance by active participants.
> >>> >>>
> >>> >>> Bronis
> >>> >>>
> >>> >>>
> >>> >>>
> >>> >>>
> >>> >>> On Wed, 26 Sep 2012, Richard Graham wrote:
> >>> >>>
> >>> >>>> Such decisions are not made lightly - I assure you this. This
> >>> >>>> was discussed at the last meeting, and this was an option
> >>> >>>> that
> >>> >>>> was not voted on. However, this was discussed again at the
> >>> >>>> end
> >>> >>>> of the meeting, and some at EuroMPI.
> >>> >>>>
> >>> >>>> Rich
> >>> >>>>
> >>> >>>> -----Original Message-----
> >>> >>>> From: mpi-forum-bounces(a)lists.mpi-forum.org
> >>> >>>> [mailto:[email protected]] On Behalf Of
> >>> >>>> Mohamad Chaarawi
> >>> >>>> Sent: Wednesday, September 26, 2012 6:30 PM
> >>> >>>> To: mpi-forum(a)lists.mpi-forum.org
> >>> >>>> Subject: Re: [Mpi-forum] Meeting schedule for 2013
> >>> >>>>
> >>> >>>> Hi Rich,
> >>> >>>>
> >>> >>>> On 9/26/2012 10:30 AM, Richard Graham wrote:
> >>> >>>>> Therefore, I propose to keep the European meeting of the
> >>> >>>>> Forum
> >>> >>>>> in conjunction with EuroMPI, and every other year also have
> >>> >>>>> a
> >>> >>>>> meeting in Japan. With this in mind, here is the proposed
> >>> >>>>> schedule for
> >>> >>>>> 2012
> >>> >>>>
> >>> >>>> The majority at the last meeting in Chicago were against
> >>> >>>> having
> >>> >>>> 2 meetings internationally. We had several votes on different
> >>> >>>> formations (4-0, 3-1, and 2-2), and 3-1 was the majority
> >>> >>>> vote.
> >>> >>>>
> >>> >>>> It would have been fair to have a similar vote in Europe and
> >>> >>>> see how the participants vote..
> >>> >>>> Did the attendance at the Vienna meeting vote on a different
> >>> >>>> schedule, or is your change coming from just some concerns
> >>> >>>> (which are also valid) from some people?
> >>> >>>>
> >>> >>>> Can we have a formal organizational Vote by eligible
> >>> >>>> organizations on this instead of just scheduling meetings
> >>> >>>> based
> >>> >>>> on suggestions that vary from meeting to meeting?
> >>> >>>> This vote can be done over the phone.
> >>> >>>>
> >>> >>>> Thanks,
> >>> >>>> Mohamad
> >>> >>>> _______________________________________________
> >>> >>>> 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
> >>> >>>>
> >>> >>> _______________________________________________
> >>> >>> mpi-forum mailing list
> >>> >>> mpi-forum(a)lists.mpi-forum.org
> >>> >>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum
> >>> >>
> >>> >> _________________________________________________________________
> >>> >> _______ Martin Schulz, schulzm(a)llnl.gov,
> >>> >> http://people.llnl.gov/schulzm CASC @ Lawrence Livermore
> >>> >> National
> >>> >> Laboratory, Livermore, USA
> >>> >>
> >>> >>
> >>> >>
> >>> >>
> >>> >> _______________________________________________
> >>> >> 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
> >>> >
> >>> >
> >>> > --
> >>> > 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. Aurélien Bouteiller
> >>> * Researcher at Innovative Computing Laboratory
> >>> * University of Tennessee
> >>> * 1122 Volunteer Boulevard, suite 309b
> >>> * Knoxville, TN 37996
> >>> * 865 974 9375
> >>> _______________________________________________
> >>> 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)
> >>
> >> _______________________________________________
> >> 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
>
> --
> * Dr. Aurélien Bouteiller
> * Researcher at Innovative Computing Laboratory
> * University of Tennessee
> * 1122 Volunteer Boulevard, suite 309b
> * Knoxville, TN 37996
> * 865 974 9375
>
>
>
>
>
>
>
>
> _______________________________________________
> 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
--
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)
