advertisement:
 Visit our sponsors

SC98 and TOP500 special November/December 1998

Primeur Live! is published during major High-Performance Computing and Networking events in Europe. This issue is published because of SC98 in Orlando and the TOP500 list published on November 1998.

Primeur is a monthly Virtual Magazine on High Performance Computing and Networking in Europe. It is produced by an editorial team composed of professionals in publishing and HPCN. You can put the editorial team to work as well. Read about our services and find out about the friendly rates.

Breaking news - just a click away:

Live issue - TOP500 analysis

On November 5, 1998 the tenth list of most powerful supercomputers in the world was published. This section gives background and analysis articles.

Check in at our analysis section for analysis of the previous lists.

Special issue -
Update on results from European R&D projects
November 30, 1998

Each year hundreds of European R&D projects produce important results. In this special issue, co-published by the magazines Primeur and VMW, we give an update from a number of projects in the HPCN and medical sectors that were supported within the Esprit or Telematics programmes from the European Commission.

This issue is published in conjunction with the large IST 1998 event (the former Esprit/Telematics conference) in Vienna.

Technology transfer

Medical Media Engineering and simulation Design and traffic Next issues are planned in April during ITIS99 (at HPCN Europe) in April and in November during IST 99 in Helsinki. When you want your project results in one of those issues, please contact the editors of Primeur magazine or Virtual Medical Worlds magazine.
Live issue - SC98 in November 1998

Primeur reports on this major supercomputing event that this year took place in Orlando Florida.

Analysis

News articles
Applications
HPCN industry
Media and visualisation
TOP500
Analysis
Metacomputing

Analysis

SC98 - Echos from SC '98: The future isn't what it used to be
SC '98, the 10-year anniversary of this conference that brings together several thousand people involved in high performance computing and communications, was of very high quality. Some impressions...
Cluster computing at SC '98
Interest in cluster computing is still growing. At the SC98 exhibition several clusters were on display in both the research and industry exhibits. One of the main trends this year is the adoption of standards in cluster programming. Interconnected PCs and workstations can be done through the use of several interconnection technologies (ATM, Gigabit Ethernet, SCI, Myrinet, Fibre Channel, HIPPI,..) Want to know more about the projects and products in cluster computing? Read the full article, that also provides many links to the sources.

Metacomputing

SC98 - ZIB Berlin linked to NLANR in long-distance supercomputing
A demonstration at the annual high-performance computing and networking conference will push the limits of visual supercomputing over global networks by computing numerical relativity theory on machines at ZIB in Berlin, in Germany and San Diego and visualizing the results in the National Laboratory for Applied Network Research (NLANR) booth on the conference floor in Orlando, Fla.
SC98 - Globus Toolkit unveiled on international computational grid
Globus, an innovative software system for building future computational grids--advanced problem-solving systems that use high-speed networks to link people with high-speed computers, databases, and other devices--will be demonstrated in Orlando by researchers from Argonne National Laboratory and the University of Southern California's Information Sciences Institute (ISI). University of Southern California's Information Sciences Institute (ISI).
SC98 - Products of GigaNet and MPI Software now compatible
GigaNet and MPI Software made their products compatible. This announcement was made in conjunction with the SC98 conference. Combining GigaNet's cLAN interconnects with MPI/Pro enables engineers to port applications to new NT systems.

Applications

SC98 - Brookhaven to showcase world's fastest non-commercial supercomputer
Just weeks after unveiling the world's fastest multipurpose non-commercial supercomputer, the team that built the 0.6-Tflop/s physics machine showcases its unique architecture and capabilities at the SC98. The team, from the U.S. Brookhaven DoE Nat. Lab. and Columbia University, will show off a sample crate of motherboards from its MPP. They will also spotlight the inexpensive 'do it yourself' construction that kept costs to around $1.8 million for the entire project. The supercomputer is a finalist for the Gordon Bell prize for price-performance at SC98.

HPCN industry

SC98 - HP displays high-performance technical-data-center solutions at SC98
Hewlett-Packard Company showcased its technical-computing scalability spectrum for high-availability, high-capacity and high-capability clustered computing solutions.
SC98 - IBM opens Advanced Computer Technology Center to support SP supercomputers
IBM is opening an Advanced Computer Technology Center that provides customers with solutions for porting and optimizing applications to IBM's RS/6000 SP. To facilitate this, the center will help independent software vendors develop programming tools and will coordinate the development of optimized software libraries.
SC98 - Objectivity ships HPSS for petabyte data retrieval
Objectivity announced the availability of its interface for the new High Performance Storage System (HPSS) hierarchical storage management (HSM) system software. HPSS is a collaborative project of government, academia, and industry, providing the capability to move very large data objects among high performance computers, workstation clusters, and storage libraries at speeds many times faster than was possible in pre-HPSS days. The product has been in testing by The European Laboratory for Particle Physics (CERN) in Switzerland and others.
SC98 - QSW developed SuperCluster products
Italian supercomputer vendor QSW developed SuperCluster products, which were shown for the first time at SC98, November 10-12,1998. The products are based on QSW's interconnect technology (QsNet) and its parallel operating environment. The SuperCluster products will be fully launched in January 1999.

Media and visualisation

SC98 - Alliance demo highlights advances in tele-immersion
The US National Computational Science Alliance has the chance to showcase a powerful new application for collaborative knowledge discovery from within a virtual environment. The application called Cave5D/Virtual Director, or Collaborative Virtual Director (CVD), combines the functionality of Cave5D, an application used to visualize and interact with large oceanographic and atmospheric datasets in the CAVE
SC98 - Silicon Graphics will jointly market CAVE systems with Pyramid
SGI intents to jointly market worldwide CAVE (Computer Assisted Virtual Environment) systems with Pyramid Systems. This new relationship represents a major step in creating worldwide standards for VR applications, specifically in advanced immersive projection systems.

Metacomputing

SC98 - Genias demonstrates latest release of Codine/GRD resource management system
Genias will demonstrate the new release of its Codine/GRD software suite for job and resource management together with Ratheyon and Compaq/Digital. The new release 4.2/1.2, includes several new options easing the management of large clusters of workstations and SMP systems: calendar management allows to automatically change job queues and user assignements during for instance the weekend; consumable resources, like software licences, tape drives and disk space can be controled to great detail; tigthly coupled MPI support has been released for MPICH.
Finistar, MPI and NCSA team to solve clustering on Windows NT
Finistar Corporation, MPI Software Technology, and the U.S. National Center for Supercomputing Applications claim to have an answer to the problem of cluster computing on Windows NT using Message Passing Interface (MPI), Virtual Interface (VI), and Fibre Channel technologies. In a demonstration, held at SC '98, cluster computing application scalability up to ten nodes was shown.

TOP500

ASCI Blue Pacific can bring IBM on first position in TOP500
The The Energy Department's Lawrence Livermore National Laboratory in California has inaugurated its new ASCI Blue Pacific supercomputer developed with IBM. The $94 million 5,856 processor machine with 3.8 Tflop/s peak, being currently the fastest machine in the world, could bring IBM to the first place of the TOP500.
Again the USA reinforces TOP500 dominance
The USA have strengthened their position in the November 5 1998 list of the worlds most powerful supercomputers. USA/Canada holds 298 of the 500 positions, as compared to 286 half a year ago. Europe is up to 120 from 114 while Japan did fall back to 69 from 83. Americans still buy USA made supercomputers, Japanese buy American and Japanese made machines, but more Japanese than US made. Europe's distribution of machines matches the sales world wide the vendors made, showing the Old Continent does not have a preference.
Energy Department's Blue Mountain goes for the top position
DoE and SGI unveiled the Blue Mountain supercomputer, located at the Department of Energy's Los Alamos National Lab. Blue Mountain is the latest advancement in the DoE's stockpile stewardship program which uses science-based methods to assess and certify the safety, security and reliability of nuclear weapons without underground nuclear testing. Blue Mountain ran the Linpack for supercomputers at a 1.6 Tflop/s, bringing them virtually on the firsts place in TOP500 list.
SC98 - IBM posts 40% growth in TOP500 list
IBM claims that it has increased its share of systems on the TOP500 list of the largest supercomputers by more than 40% in just the past six months. IBM systems now account for 105 of the world's fastest computers, an increase of 30 installations since June

SC98 - Sun Starfire fast mid-range machine
With 4 clusters of 64 processors, the SUN 10000 Starfire reaches a 123.9 Tflop/s on the Linpack benchmark. Sun claims this is the fastest perfomance in its class on the benchmark used in the TOP500. With two clusters it delivers 66.9 Gflop/s . The Sun HPC 10000 server is currently the leading architecture on the TOP500.
SC98 - SUN firmly in second position in TOP500 list
One year after its first entry in the TOP500 list, Sun now firmly holds the second position measured in number of machines after SGI/Cray. Sun servers accounted for the second highest number of total systems of all companies on the list, accounting for 126 systems -- a 14% increase from the June 1998 Top500 list.
Germany and UK dominate European TOP10 machines
According to the new TOP500 list, the T3E900 with 876 processors at UK Meteorological Office is still the largest installation in Europe and on rank5 world wide. The second machine in Europe is the T3E1200 with 612 processors in Manchester, the new UK scientific supercomputer. The German sites Deutscher Wetterdienst, Max Planck, HWW Stuttgart and FZJ Juelich follow before the next UK machine: EPCC in Edinburgh. These are all SGI T3E computers. On place 8 is the first non-SGI machine in Europe's TOP10: the VPP700 at ECMWF. A second T3E in Juelich is in on place nine. CEA in France on position 10 closes the TOP10.

SC98 - Stay at home and still be there with the live webcast
Live webcasts of special portions of SC98 were broadcasted over the World Wide Web (WWW) this year for the first time ever! NCSA (National Center for Supercomputing Applications) offered live feeds from the Exhibit floor, and selected sessions on Tuesday, Wednesday, and Thursday.
SC98 - SCinet98 at 2,5 Gbps
If you want to bring the world's leading experts in supercomputing and high performance networking together, it helps to move quickly. Or at least move data quickly.At SC98 a specially-created high performance communications network will be running an OC-48 Local Area Network (LAN). Called SCinet (shorthand for Scientific Computing Network) the on-site network is ready for implementation. The all-volunteer SCinet98 staff has designed a world-class network to support the networking requirements of show participants, including on-site demonstrations and research. The SCinet98 LAN is built upon an OC-48 (2.5 Gbps) Asynchronous Transfer Mode (ATM) backbone.
SC98 - SCinet98 is the World's First Show Network with an OC-48c ATM Backbone
Th ATM (asynchronous transfer mode) network backbone at SuperComputing '98 (SC98), the High Performance Networking and Computing Conference, being held this week in Orlando, Florida. For the third consecutive year, SC98 will rely on FORE solutions to offer tradeshow exhibitors ATM and switched Ethernet connections, providing high-performance connectivity to many of the world's largest ATM networks.
SC98 - HP and Seagate plan to produce 4 Gbit/s fibre channel networking products
Hewlett-Packard Company and Seagate Technology plan to deliver products based on 2 Gbit/sec Fibre Channel technology, with a roadmap leading to 4 Gbit and higher. At Supercomputing '98 Hewlett-Packard Company also showcased its 64-bit PA-8500 processor.
SC98 - European networking at SC98
At SC98 High Performance Networking and Computing Conference in Orlando, Florida, November 7 to 13, the European networking demonstrations booth demonstrated research and development, as well as best practice and demonstration projects being conducted with European Community support. It is coordinated by the NETAPDEX project, which promotes network usage in Europe.