[Beowulf] how can I calculate peak performance of a cluster
Robert G. Brown
rgb at phy.duke.edu
Tue Sep 27 07:44:41 PDT 2005
guren at be.itu.edu.tr writes:
> Hi,
>
> I want to calculate peak performance of cluster which HP HPC Cluster
> consist of different numbers of INTEL Xeon 3.4GHz EM64T , INTEL Xeon DP
> 3.2GHz EM64T ,INTEL MP XEON 3.16GHz EM64T processors. I think it is easy,
> but I couldn't find.many
There are lots of choices. lmbench, spec (if your organization
participates), linpack, hpcc, benchmaster, stream, netpipe. Google Is
Your Friend (GIYF) -- use these as keyworks in searches.
Be warned -- using a benchmark to "calculate peak performance" is
usually a waste of time and is generally done for political reasons, not
practical/operational ones. You'd be much better off using the
microbenchmarks or macrobenchmarks on this list to develop a performance
PROFILE of your cluster nodes -- learning how fast they each can access
streaming memory (stream), access memory randomly or do transcendental
function evals (benchmaster), run representative applications (spec), or
do various microscopic operations (lmbench, netpipe, benchmaster). That
kind of information is actually useful both in selecting a cluster
architecture that matches your application's needs and in tuning your
application to your cluster architecture.
rgb
>
> many thanks for help
>
> _______________________________________________
> Beowulf mailing list, Beowulf at beowulf.org
> To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
URL: <http://www.beowulf.org/pipermail/beowulf/attachments/20050927/d0d2d7fc/attachment.sig>
More information about the Beowulf
mailing list