alfred The Great New IBM Linux Cluster

alfred, the great new IBM Cluster

Advanced Research Computing has taken delivery of our new system, an IBM eServer 1350. This is a Linux Cluster system consisting of 48 compute nodes, a head node, a storage node and a management node.

The configuration is a follows:

Compute Nodes:	dual 2.4 GHz Xeon	1 GByte RAM	40 GByte Disc (10 GB /scratch)
Head Node:	dual 2.4 GHz Xeon	1 GByte RAM	36 GByte RAID1 (mirrored)
Storage node:	dual 2.4 GHz Xeon	1 GByte RAM	2.4 TByte Disc (= 2 TByte RAID5)
Management Node:	dual 2.4 GHz Xeon	1 GByte RAM	35 GByte RAID1

All the nodes are connected to a 1 Gbps Ethernet switch, and we have a 40 node Myrinet switch connecting the nodes in the parallel partitions.

Usage

1. Connection
   Connections must be made using ssh. (PuTTy supports ssh...).
   Connect to alfred.hpc.unimelb.edu.au
   Jobs can be also be submitted over the Grid. If you wish to do this please arrange to see dirk and discuss how.
2. Basic job submission
   The alfred usage page gives a quick overview of how to get started using the system.
3. Software
   The alfred software page has a list of software currently available, and links to documentation. It is not exhaustive!

Availability

alfred is available to all researchers at the University of Melbourne. To get an account you need to fill out an online form (click here) which is automatically mailed to the Advanced Research Computing staff. They usually will ask to interview you to confirm your exact requirements, check that we can meet them etc. You will then be asked to fill out a paper form, (Order for Computing Services), and an account will be created.

Pictures

The normal view doors closed

doors off access to the console

A fisheye view

Using the console

Created: 23rd April 2003	Maintainer: Research Computing Services Information Services The University of Melbourne Disclaimer
Last modified: Tuesday, 09-Jan-2007 12:36:35 EST Access: Unrestricted Copyright © Sunday, 08-Nov-2009 01:47:17 EST The University of Melbourne.