Pete was divided into two sub-clusters, one for EAS and one for NCN. The difference was mainly the amount of memory per node.
|Sub-Cluster||Number of Nodes||Processors per Node||Cores per Node||Memory per Node||Interconnect||TeraFlops|
|Pete-EAS||84||Two 2.33 GHz Dual-Core Intel Xeon E5140||4||8 GB||Gigabit Ethernet||3.13|
|Pete-NCN||82||Two 2.33 GHz Dual-Core Intel Xeon E5140||4||16 GB||Gigabit Ethernet||3.06|
Pete also featured a 17 TB NFS scratch filesystem and a 16 TB Lustre scratch filesystem.
All Pete nodes ran Red Hat Enterprise Linux 4 (RHEL4) and used PBSPro 9.x for resource and job management. Pete also ran jobs for BoilerGrid whenever processors in it would otherwise have been idle.