compute resources
the luhpcc maintains a 240 compute core linux cluster named wesley that was installed in the summer of 2011.
system name: wesley.lakeheadu.ca
operating system: centos 5.5
number of nodes: 20
total number of processing cores: 240
number of processing cores per node: 12
number of fermi gpus: 1
interconnection type: 1 gb ethernet
ram per node: 16 @ 48 gb, 4 @ 96 gb
ram per core: 192 @ 4 gb, 48 @ 8 gb
permanent shared (nfs) disk space: 12 tb
scratch disk space per node: 1 tb (2x500 gb in raid 0)
batch facility/scheduler: torque and maui
wesley is composed of 20 compute nodes housed within five dell poweredge c6100 series chassis (4 nodes/chassis), a poweredge r710 login/head node, and a poweredge r710 storage node with 12 tb of direct attached storage (das) in a powervault md 1200. a powerconnect 6248 gigabit ethernet switch serves as an interconnection fabric between all of these components making wesley suitable for serial farming and low-bandwidth mpi cross-node calculations. a poweredge c410x pcie expansion chassis provides general purpose graphics processing unit (gpgpu) capabilities. the c410x can hold up to 16 gpgpus (currently a single nvidia tesla m2050 gpgpu is installed).
each compute node contains two intel xeon x5650 hex-core processors @ 2.66 ghz and two 500 gb sata drives in a hardware raid 0 configuration (for 1 tb capacity and higher throughput). sixteen of the twenty nodes have 48 gb of ram. the remaining four have 96 gb (wes-04-00 through wes-04-03).
the login and storage nodes contain two intel xeon x5650 hex-core processors @ 2.66 ghz, mirrored 146 gb sas operating system drives, and 24 gb of ram.
