[mpi3-coll] July/August telecon
siebert at it.neclab.eu
Thu Jul 31 05:11:50 CDT 2008
Torsten Hoefler wrote:
> On Mon, Jul 28, 2008 at 10:13:49AM -0400, Torsten Hoefler wrote:
>> 4) Updates on MPI Plans (I have a slightly different
>> proposal/implementation than Christian for the same thing)
> I just added my proposal to the wiki in case anybody wants to take a
> look at it before or during the telecon:
> (scroll down to the bottom).
And for completeness, I attached my interface proposal for "MPI plans"
to the same wiki page (very bottom). Alternatively the direct link is:
This description has already been revised a couple of times, and is
since 10th of July in a quite stable and hopefully complete state. A
second implementation (independent of Torsten's
reference-implementation) for NEC's proprietary MPI has been completed
and is currently evaluated/tuned in application contexts.
The difference to Torsten's proposal is quite small and is mainly based
on the question if we should give some memory management things to the
programmer (because she might know already how large the plan will
become) or hide all these aspects. I'm locking forward to discuss the
advantages/disadvantages of both ways with the audience.
PS: Since April there is also a first draft for persistent collectives
in the wiki (see appropriate section). The direct link is
Christian Siebert, Dipl.-Inf. Research Associate
NEC Laboratories Europe, NEC Europe Ltd.
Rathausallee 10, D-53757 Sankt Augustin, Germany
Phone: +49 (0) 2241 - 92 52 44 Fax: +49 (0) 2241 - 92 52 99
(Registered Office: 1 Victoria Road, London W3 6BL, 2832014)
More information about the mpiwg-coll