[Beowulf] Using Autoparallel compilers or Multi-Threaded librarieswith MPI

[Tom Elken]

 > -----Original Message-----
> [mailto:beowulf-bounces at beowulf.org] On Behalf Of Tom Elken
> Sent: Thursday, November 29, 2007 11:27 AM
> 
> Have you used compiler auto-parallel features mixed with MPI with
> success on your clusters?
> 
> Have you used multi-threaded math or scientific libraries 
> mixed with MPI
> with success on your clusters?
> 
> If you just want to 'reply' to me only with simpler Yes/No answers, I
> will report on a summary of the results to this list and to 
> the SPEC HPG committee.

Results of the VERY non-scientific survey:

# reporting use of Autoparallel features with MPI:          0

# reporting use of multi-threaded math libraries with MPI:  1

The '1' was using multithreaded BLAS and MPI on HPL (a benchmark, not an
application) and his recollection was that it was not a win over pure
MPI.

I'll let you know later when we've resolved the discussion in the HPG
committee on how this might affect SPEC MPI2007.

But the discussion engendered by this post (and similar ones) was quite
entertaining and educational, with a lot of heat and some light.
Especially between my ex-SGI colleague, Joe Landman, and my ex-PathScale
colleague, Greg Lindahl.

Thanks,
Tom

> 
> If you have success or failure stories that might be useful to the
> Beowulf list, please 'reply-all'.  
> 
> Thanks,
> Tom Elken,
> member SPEC HPG committee
> -----------------------------
> *  For example, if an autoparallelizing compiler could find effective
> 4-way thread-level parallelism in an MPI code and you were 
> running on a
> cluster of 8 nodes each with two quad-core CPUs, 64 cores total, you
> might choose to run with 16 MPI threads and set your NUM_THREADS
> variable to 4, to run with all 64 cores of the cluster executing work
> with reasonable efficiency. 
> 
> _______________________________________________
> Beowulf mailing list, Beowulf at beowulf.org
> To change your subscription (digest mode or unsubscribe) 
> visit http://www.beowulf.org/mailman/listinfo/beowulf
>