[Beowulf] benchmarking mpich

[Gabb, Henry]

Hi Deek,
The Intel MPI Benchmarks are designed to measure the performance of MPI
functions. For example, if you want to compare the barrier
implementation from different MPI's, these benchmarks could help you. It
should work with any MPI, including MPICH. You can download it here:
http://www.intel.com/cd/software/products/asmo-na/eng/cluster/clustertoo
lkit/219848.htm

Best regards,

Henry A. Gabb
Intel Parallel Applications Center
1906 Fox Drive
Champaign, IL 61820

Phone: (217) 403-4212   Fax: (217) 403-4399
E-mail: henry.gabb at intel.com


> Message: 4
> Date: Thu, 19 Jan 2006 13:46:44 -0800
> From: decon brody <decon.brody at gmail.com>
> Subject: [Beowulf] benchmarking mpich
> To: beowulf at beowulf.org
> Message-ID:
> 	<c437a1580601191346k6a61661ftc1ac7b93418f72bb at mail.gmail.com>
> Content-Type: text/plain; charset="iso-8859-1"
> 
> Hello all,
> 
> I'm looking for a way to find somekind of benchmarking number for
mpi(ch).
> I'm not sure what I'm looking for, perhaps just a time value that's
> repeatable and scalable.  I already run linpack which of course gives
me
> numbers relevant to the whole cluster but not specific to mpi.
> 
> Any suggestions would be greatly appriciated!
> 
> -deek