Hi everybody,
The decision to form the MPI subsetting working group is based on the
informal vote that was held at the previous meeting. A lot of people
raised their hands then.
I would suggest that if you're interested in this topic, please
subscribe to the
http://lists.mpi-forum.org/mailman/listinfo.cgi/Mpi3-subsetting
right away. By early next week I'm going to drop a message into that
list to kick off the discussion of the scope and logistics.
Best regards.
Alexander
--
Dr Alexander Supalov
Intel GmbH
Hermuelheimer Strasse 8a
50321 Bruehl, Germany
Phone: +49 2232 209034
Mobile: +49 173 511 8735
Fax: +49 2232 209029
---------------------------------------------------------------------
Intel GmbH
Dornacher Strasse 1
85622 Feldkirchen/Muenchen Germany
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 (BLZ 502 109 00) 600119052
This e-mail and any attachments may contain confidential material for
the sole use of the intended recipient(s). Any review or distribution
by others is strictly prohibited. If you are not the intended
recipient, please contact the sender and delete all copies.
I apologize for the delay; I had budgeted some time immediately
following the January meeting for collating and publishing all my
notes, but the web site was not available at that time. So the
publishing step got put off until I had some more cycles available --
which finally happened late last week. Secretary notes from the first
meeting are [finally] available here:
http://meetings.mpi-forum.org/secretary/
I think Rich will be adding a link to that page from the main meetings
web site somewhere, too.
Note that these are notes from *before and during the January
meeting*; I made no attempt to update the information with what has
happened since the meeting.
--
Jeff Squyres
Cisco Systems
Since our first meeting in Chicago several people have been looking at
options for alternative meeting locations that would reduce the meeting
costs, and make it easier for people from Asia to attend these meetings. NO
change is happening in the March meeting, it will still take place at the
Chicago Marriott O'Hare, with the meeting rate being $175/person. The
following meetings will alternate between Chicago, IL and Menlo Park, CA.
Microsoft will host the meetings in Chicago, when their downtown facility is
available (or else we will meet at the Marriott), and Sun will host the
meetings in CA.
So the plan would be
April 28-30, 2008 Microsoft facility in Chicago
June 30 - July 2nd, Menlo Park
Sept 3-5, Dublin Ireland
Oct 20-22, Chicago
Dec 15-17, Menlo Park
If in doubt, https://meetings.mpi-forum.org/Meeting_details.html is kept up
to date with meeting information.
Feel free to suggest alternatives that meet the objectives of reducing the
meeting cost and making it easier for the Asian community to participate in
this. Barring any such suggestions this will become the plan of record on
Wed, 2/13, to make it possible for people to purchase airplane tickets to
the March meeting.
Thanks,
Rich
Can someone take a look at the second example in:
http://www.mpi-forum.org/docs/mpi-20-html/node124.htm
and also to the example in:
http://www.mpi-forum.org/docs/mpi-20-html/node125.htm
I believe in both cases the lines saying
j = map(i)/p
k = MOD(map(i),p)
should read
j = map(i)/m
k = MOD(map(i),m)
Additionally, in the first example again in:
http://www.mpi-forum.org/docs/mpi-20-html/node124.htm
in the part where origin and target indices are computed, there are
two lines reading:
oindex(total(j) + count(j)) = i
tindex(total(j) + count(j)) = k
Should'nt it read like this (for indices being zero-based)?
oindex(total(j) + count(j)) = i - 1
tindex(total(j) + count(j)) = k - 1
Regards,
--
Lisandro Dalc�n
---------------
Centro Internacional de M�todos Computacionales en Ingenier�a (CIMEC)
Instituto de Desarrollo Tecnol�gico para la Industria Qu�mica (INTEC)
Consejo Nacional de Investigaciones Cient�ficas y T�cnicas (CONICET)
PTLC - G�emes 3450, (3000) Santa Fe, Argentina
Tel/Fax: +54-(0)342-451.1594
It turns out that if we want to get a room block at the hotel that is with
in walking distance from the meeting location we really need to get a head
count as soon as possible, as the hotel is making it more
difficult/expensive to make such reservations with time. I will go with the
head count I get by mid-day Friday, 2/8/2008. Also, most folks who have
responded have not indicated if they will need a room starting on the 2nd,
or the 3rd, so I will just ask for the group rate starting on 9/3. This is
just for a group rate on the hotel ...
Thanks,
Rich
On 2/6/08 1:34 PM, "Richard Graham" <rlgraham(a)ornl.gov> wrote:
>
> ------ Forwarded Message
> From: Richard Graham <rlgraham(a)ornl.gov>
> Date: Wed, 06 Feb 2008 12:28:25 -0500
> To: <mpi-forum(a)lists.mpi-forum.org>
> Conversation: Dublin MPI Forum
> Subject: Dublin MPI Forum
>
> The organizers of the Euro PVM/MPI meeting, who will be hosting the MPI
> Forum in Dublin, are willing to negotiate a block of rooms for the Sept 3-5
> meeting. To do this, they need to have an indication of what the demand
> will be for the hotel rooms. What I need to know is who would like to take
> advantage of the group room rate, if you would plan to arrive on Sept 2nd,
> or Sept. 3rd, and if you plan to attend the Euro PVM/MPI meeting. If you
> would like to take advantage of the group room rate, please send me a note
> (rlgraham(a)ornl.gov)
>
> Thanks,
> Rich
>
> ------ End of Forwarded Message
>
> _______________________________________________
> mpi-forum mailing list
> mpi-forum(a)cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/mpi-forum
On
http://www.cs.uiuc.edu/homes/wgropp/projects/parallel/MPI/mpi-errata/index.…
we have sorted out which topic on the errata list fits to which
MPI 2.1 / 2.2 / 3.0 slot, and in case of 3.0 to which subgroup and
mail reflector.
(You can also see, that nearly all MPI 2.1 Ballot 4 topics have a full
proposal text on thee web now.)
Please, check for all red-marked topics to help to get them into
the right position.
If a 3.0 subgroup/mail reflector is set up, please check that
this is included on the errata page and the RED flag is removed.
Sometimes, there is an open question, but no yet any full proposal
to solve a problem.
Best regards
Rolf
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)
Dear MPI Forum,
we want to publically announce the MPI-3 Collectives workgroup. This
email will be the only mail regarding the workgroup that will be sent to
the MPI Forum mailinglist. All followup discussions should be directed
to the mpi3-coll [1] mailinglist.
The current topics of the workingroup include:
1) non-blocking collective operations
2) topological collective operations (associated with topological
communicators)
3) sparse collective communication (extended definition of parnership
relations and communicated data)
4) new collective operations (such as MPI_Reduce_scatter_block and
MPI_Exscan_all proposed by Jesper)
Further topic proposals are welcome but we would like to see a full
proposal that is written MPI-like and can be discussed. Other
discussions might be held on the Mailinglist/Teleconferences/Wiki but
will not have proposal status (i.e., will not be presented to the full
Forum) unless the requirements mentioned above are achieved (and the
workgroup voted positively).
Policies: we use a simple majority straw-man vote per (registered)
attendee held online (note: those votes are not ratified by the
forum and will only be valid in the subgroup). Our proposals will only
be accepted by the MPI Forum if an open-source implementation is
available.
All proposals and votes will be documented in the Wiki (URL to be
defined).
Our first teleconference will be held on February 12th at noon EST. The
call-in number will be announced on the mailinglist.
Best Regards,
Andrew Lumsdaine and Torsten Hoefler
[1]: http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-coll
--
bash$ :(){ :|:&};: --------------------- http://www.unixer.de/ -----
Indiana University | http://www.indiana.edu
Open Systems Lab | http://osl.iu.edu/
150 S. Woodlawn Ave. | Bloomington, IN, 474045-7104 | USA
Lindley Hall Room 135 | +01 (812) 855-3608
