I have read the slides and have some comments on the issues related to
MPI_INIT time assertions (slide 4).
1) I agree the set of assertions must be negotiated but unanimous
implementor acceptance should not be needed. Defining a assertion in the
standard that does not help a particular MPI implementation is pretty
harmless. Defining one that does not help any implementation is useless.
Implementors are in a good position to recognize assertions that can never
be useful and help kill them but no implementor should have veto power..
2) I have not heard a decent rationale for implementation specific
assertions. Assertions are not about the implementation. They are about a
particular application and its requirements in relation to the MPI
standard. The MPI implementation simply ignores any assertion that is not
useful to optimizing that implementation.
3)
a) Any tool that intercepts variants of MPI_INIT can remove assertions
it does not like. Removing an assertion can never make an application quit
working properly.
b) I am not sure what "Layered underlying libraries" means but the
observation on the slide does apply to what I would call "third party
libraries". If some library that will be linked into an application (say
a parallel math library) makes calls to MPI_Cancel for ISEND requests than
that library should check that the MPI_NO_SEND_CANCEL assertion was not
used and bail out if it was.
Dick
Dick Treumann - MPI Team/TCEM
IBM Systems & Technology Group
Dept 0lva / MS P963 -- 2455 South Road -- Poughkeepsie, NY 12601
Tele (845) 433-7846 Fax (845) 433-8363
"Supalov,
Alexander"
<alexander.supalo To
v_at_[hidden]> "MPI 3.0 Sub-setting working group"
Sent by: <mpi3-subsetting_at_lists.mpi-forum.or
mpi3-subsetting-b g>
ounces_at_lists.mpi- cc
forum.org
Subject
Re: [Mpi3-subsetting] call for a
06/27/2008 08:55 report-out and session chair
AM
Please respond to
"MPI 3.0
Sub-setting
working group"
<mpi3-subsetting@
lists.mpi-forum.o
rg>
Hi everybody,
Rich Graham kindly agreed to chair the subsetting meeting for us.
A draft report-out can be found in
http://svn.mpi-forum.org/trac/mpi-forum-web/attachment/wiki/SubsettingWikiPage/MPI-3%20subsetting%2C%20take%203.pdf
. I have to use a change-protected PDF for this external publication.
Please comment.
I've also added last meeting minutes, and am setting about updating the
proposal accordingly now.
Best regards.
Alexander
From: Supalov, Alexander
Sent: Thursday, June 26, 2008 5:36 PM
To: 'MPI 3.0 Sub-setting working group'
Subject: call for a report-out and session chair
Hi everybody,
As I won't be at the meeting next week, we need a volunteer to report out
to the Forum and then chair the subsetting WG meeting 5:15 pm - 7:00 pm on
Tuesday, June 1. Who wants to do this? Please let me know.
In the meantime, I'll prepare and upload a draft report-out, and update the
current proposal using the materials of our recent meeting by EOW.
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.
_______________________________________________
mpi3-subsetting mailing list
mpi3-subsetting_at_[hidden]
http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-subsetting
|
![[Flag]](../../../../../images/flag-us.gif){kind=small}