IT Department

The primary task of the IT department is to provide a well-functioning computer infrastructure to all employees of the Institute, so that the scientific work is optimally supported. Service includes giving direct support in case of usage problems and supporting the scientists and the members of the service departments in solving more specialized IT problems. For solving computer-intensive problems such as simulations and data analysis, the IT department runs two high-performance compute clusters. The IT department also provides the means for videoconferencing, which enables members of the AEI to easily join meetings in different locations of the world. Particularly important are the meetings held between the two AEI-sites, Golm and Hannover.

High performance compute clusters

In Potsdam the AEI runs two kinds of high performance compute clusters. One type is for general purpose if a complex calculation is needed, but is mainly used by the Numerical Relativity group to perform highly compute intensive simulations. The other kind of cluster is used for data analysis, which is of interest of the Gravitational waves group.

General purpose high performance compute clusters (Peyote, Belladonna, Damiana)

A special characteristic for clusters doing simulations is the high speed interconnect networks. At present the AEI has three generations of cluster. The cluster Peyote was purchased in 2003 and has grown in three phases to a cluster with 178 Intel XEON compute nodes. The high speed network is based on Gigabit Ethernet. In 2006 Belladonna was added. Belladonna has 64 compute nodes with each two AMD Opteron processors (4 cores). Belladonna has two high speed networks, one is Infiniband and one is Gigabit Ethernet. In production the Infiniband network is mostly used to shorten the execution time of the scientific runs. The third cluster is Damiana, which was installed begin 2007. It consists of 168 INTEL Woodcrest compute nodes and uses Infiniband as high speed interconnect network. This system could make it to rank 192 in the Top 500 list of 2007. More details about the clusters named above can be found on the cluster WEB pages. Look at Cactus for more information of the programming environment.

High performance data analysis compute cluster (MERLIN/MORGANE)

The characteristic of the cluster for data analysis is the huge amount of data that needs to be stored and processed. To manage the analysis of the huge amounts of data gathered by the gravitational wave detector GEO600, the cluster dubbed "Merlin" was installed in December 2002. After almost 6 years of operation, with about 120 of the initial 180 nodes still operational, the cluster was shut down for energy efficiency reasons.

In 2007 new hardware was installed. The new part of the "Merlin/Morgane dual cluster" consists of 615 compute nodes, each with dual AMD Opteron CPUs. While the cluster in its beginnings was reserved exclusively for data analysis by the GEO group of the institute, it has been opened for a wider public during the last year. For further information, please look at the resources web pages of the the gravitational waves group.

Computer infrastructure

Video conference system

Important documents

Talks

IT department staff