PRESS RELEASE

DAS - "High Performance Server tailored to your needs"

Dutch Research School Orders Pentium Pro Based High Performance
Multi-Processor System From Parsytec

ASCI (Advanced School for Computing and Imaging), founded by four Dutch
universities, received an equipment grant from SION, the computer science
foundation of the Netherlands Organization for Scientific Research (NWO) to
fund the Distributed ASCI Supercomputer (DAS) project. In January  97
Parsytec was chosen as the supplier as it has many years of experience of
building large parallel computing systems. The company is also a leading
supplier of industrial high performance image recognition systems for
surface inspection and automatic data capture. The hardware platform of
these systems is based on the Parsytec CC architecture and it is exactly
this system technology which Parsytec also supplies to its research
customers. It facilitates straight-forward integration of standard PC
components and the application of industrial quality standards even into
supercomputer systems.
The DAS system consists of a Parsytec CCi with 144 nodes, each of which
contains a standard 200 MHz Pentium Pro processor, 64 MB RAM, and a 2.5 GB
hard disk. The nodes are connected by Myrinet, a 2.4 gigabit communication
technology supplied by Myricom, a California company founded by former
CalTech professor Charles Seitz and his associates. The goal of this project
is the construction of a high performance parallel and distributed system
built from standard components commercially available off-the-shelf. Total
system performance will be 29 GFLOPS or 87 GOPS and will thus represent one
of the largest multicomputer systems in the Netherlands. A further
exceptional system feature is that it will be distributed over four
different sites which will all be linked up to each other via ATM networks.

In 1993, four Dutch universities, the Technische Universiteit Delft (TUD),
the Universiteit van Amsterdam, the Vrije Universiteit, and the
Rijksuniversiteit Leiden founded ASCI (Advanced School for Computing and
Imaging). In the DAS project the system with a total of 144 Pentium Pro
processors (frequency 200 MHz) will be distributed over this four sites and
linked up to each other via high performance ATM networks. For the internal
communication between the processors in each system the Myrinet technology
will be used. So apart from the HS (High Speed) link which was developed by
Parsytec there is now a choice of a second communication technology
available for the CC. As the operating system


the DAS system will run UNIX.
The universities of Amsterdam, Delft and Leiden will each receive a 26 node
cluster, the Vrije Universiteit will receive a 66 node system. At each site,
one node will be a file server and one will be a gateway to the other sites.
The remaining 24 or 64 nodes will be computing nodes. Although not a formal
member of ASCI, the University of Utrecht will also participate in this
project.

Professor Andy S. Tanenbaum, Scientific Director of ASCI, professor of
computer science at the Vrije Universiteit in Amsterdam and one of the main
scientific co-ordinators of the DAS project, explains the main aims: "We
want to show that we are now able to build a supercomputer from
off-the-shelf PC components and which can even meet industrial quality
standards." Further project goals are the research of these parallel
supercomputer systems and the development of application software for
different HPC areas and image processing applications.

By selecting Parsytec ASCI has gained a project partner with many years of
experience of working in the field of parallel processing technology and
which also has extensive know-how of integrating PC standards under
industrial quality aspects. The CC constitutes the hardware platform of
Parsytec's high speed image recognition system which is used for industrial
surface inspection and automatic data capture under high performance and
reliability demands. It is exactly these system features which are
imperative for the system's industrial use and which were of prime
importance when selecting the technology for the DAS system. A further
reason for selecting the CC technology is Parsytec's development of
intelligent mechanisms which facilitate the control and operation of the
entire computer system from a central location. Parsytec also gains know-how
from the DAS project, i.e. the short term integration of the latest
technologies such as Myrinet as a communication technology and the
availability of improved system software. Over and above this, key
technologies emerge as a result of the project work such as special
application software in the imaging field. This is of utmost importance for
Parsytec for the development of new high performance inspection systems.

4200 characters					24th February 1997

For further information:
Parsytec Computer GmbH
Thomas Glodek/Alexandra Reichel		Tel. +49 241 9696-200
Public Relations				Fax +49 241 9696-500
Auf der Huels 183				CIS: 100074,2715
D-52068 Aachen				http://www.parsytec.de
Germany