direkt zum Inhalt springen

direkt zum Hauptnavigationsmenü

Sie sind hier

TU Berlin

Page Content

TUSCI – TU Berlin SCI Cluster

Lupe [1]

In order to support our research and teaching activities in the fields of cluster computing, application oriented resource management, grid computing, parallel operating systems, parallel simulations and parallel programming we operate the TU Berlin SCI cluster system (TUSCI). The cluster consists of 16 nodes with 2 Intel P4 Xeon processors and 1GB main memory each. The nodes are connected using the high performance interconnect SCI (Scalable Coherent Interface) building a 2D torus topology.

Lupe [2]

The SCI high performance interconnect provides very low latency and support for hardware based distributed shared memory (DSM). Implementing SCI as the interconnect of our cluster system achieves much more flexibility than provided by message passing support of other high performance networks.

Managed by the computing center software CCS [3] the cluster systems runs the Linux operation system. While implementing network adapter cards and drivers from Dolphin [4]; message passing support is provides by the MPIch suite from RWTH Aachen [5].

Lupe [6]

Today, the TUSCI cluster system primary serves as base for our projects in the field of grid computing, such as the Virtual Resource Manager (VRM), the Workflow Online Resource Management (WORM) Systems. Additionally the cluster is part of the Berlin Brandenburg Grid environment (BBGrid).


Jan Schoenherr
------ Links: ------

Zusatzinformationen / Extras

Quick Access:

Schnellnavigation zur Seite über Nummerneingabe

Copyright TU Berlin 2008