Gareth.Williams at csiro.au Gareth.Williams at csiro.au
Tue Mar 3 18:26:39 MST 2009

> -----Original Message-----
> From: Arnau Bria [mailto:arnaubria at pic.es]
> Sent: Wednesday, 4 March 2009 2:21 AM
> To: mauiusers at supercluster.org
> Subject: [Mauiusers] DEDICATEDPS vs DEDICATEDPES
> Hi all,
> Could someone help me to undestand teh diff bewteen DEDICATEDPES and
> processor-equivalent second means normalized cpu time?
> TIA,
> Arnau

Hi Arnau,

PE is a measure of how many processors a job is preventing others from accessing.  Lets say you have a cluster of dual socket, 4 core machines with 16GB per node.

A job that requests 1 core and 8GB of memory is effectively using half of a node (if we expect most/average jobs to request about 2GB of memory per core requested) so has a PE count of 4.
A job that requests 16GB is effectively stopping any other jobs from running on a node so PE=8

The concept breaks down on a system which doesn't have highly uniform nodes.

>From http://www.clusterresources.com/products/maui/docs/5.1.2priorityfactors.shtml
: The final resource component, PE, represents 'Processor Equivalents'.  This component can be viewed as a processor-weighted maximum 'percentage of total resources' factor.  For example, if a job requested 25% of the processors and 50% of the total memory on a 128 processor O2K system, it would have a PE value of MAX(25,50) * 128, or 64.  The concept of PE's may be a little awkward to grasp initially but it is a highly effective metric in shared resource systems.

-- Gareth

More information about the mauiusers mailing list