[mpich-discuss] Why do predefined MPI_Ops function elementwise in MPI_Accumulate, but not in MPI-1 routines?

[Jeff Hammond]

I think user-defined ops in MPI_Accumulate are a hard sell.  Fixing
MPI_Reduce to support things that MPI_Accumulate does seems like a
much easier task from an MPI standard perspective.  However, since I
have heard nothing about a timeline for MPI 3.1 and this won't make it
into MPI 3.0, it would be really nice to have
MPIX_Reduce_like_accumulate in MPICH2.

Jeff

On Tue, Apr 24, 2012 at 2:19 PM, Jed Brown <jedbrown at mcs.anl.gov> wrote:
> On Tue, Apr 24, 2012 at 14:09, Dave Goodell <goodell at mcs.anl.gov> wrote:
>>
>> Winking-face aside, the solution that I think you really want must come
>> from the MPI Forum, not MPICH2.  Even if we (MPICH2) choose to support the
>> extensions of (a), (b), and/or (c), you'll still be missing them when you
>> use a non-MPICH2 MPI implementation.  And I wouldn't be surprised if some
>> downstream vendors objected to (a) and wouldn't support it because of an
>> implementation impact.
>
>
> Yeah, I'm aware of this. I'm under the impression that Torsten, Jeff, and
> others have advocated (a) at the Forum in the past, and that it has not yet
> been accepted or rejected.
>
> _______________________________________________
> mpich-discuss mailing list     mpich-discuss at 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 at alcf.anl.gov / (630) 252-5381
http://www.linkedin.com/in/jeffhammond
https://wiki.alcf.anl.gov/parts/index.php/User:Jhammond (in-progress)
https://wiki.alcf.anl.gov/old/index.php/User:Jhammond (deprecated)
https://wiki-old.alcf.anl.gov/index.php/User:Jhammond(deprecated)