[mpich-discuss] Benchmarks Using Allgather

[Jeff Hammond]

Can you describe your optimizations in some more detail?  I know a few
codes that use MPI_Allgather often but in different ways.  Will the
improvement be in latency or bandwidth?  Are they scalable to >100,000
ranks?

Not all parallel matrix multiplication algorithms use MPI_Allgather,
by the way.  The best I have seen lately uses mostly MPI_Bcast.

Jeff

On Mon, Jul 11, 2011 at 12:57 PM, vaibhav dutt
<vaibhavsupersaiyan9 at gmail.com> wrote:
> Hi,
> I know this may not be the best place to ask this question but since there
> are so many experts
> on this mailing list, I thought I would get some good help.
> I have made some optimization in the MPI_Allgather collective and I want to
> test it on some user
> application  that makes use of this collective quite often. I just know that
> this collective is used
> in matrix multiplication operations but I do not know of any
> benchmark/parallel application
> that uses it heavily or even moderately. Can anybody suggest me some
> applications
> that use MPI_Allgather quite often?
> Thanks.
> _______________________________________________
> mpich-discuss mailing list
> mpich-discuss at mcs.anl.gov
> 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/index.php/User:Jhammond