Press conference

>PetaFlop computing requires the softron for better software design productivity instead of increased hardware performance
>With Thunder, Quadrics continues to drive Linux cluster's performance over the edge

>Applications

>Simulating the birth of the Universe to understand its present-day growth of structure
>BP uses HPC power for seismic imaging
>HPC power used in physical infrastructure assessment and protection against natural and human disasters
>SAP Business Solution to convert to adaptive computing

>Company news

>Cray X1 supercomputer processors again are most powerful on TOP500 list
>Partial "Blue Gene" systems are now two of the Top Ten most powerful supercomputers on Earth
>PathScale and Absoft collaboration
>AMD Opteron processor-based installations see sevenfold increase in TOP500 Supercomputer list

Press conference

Heidelberg 23 June 2004 No time for the "big lunch" but sandwiches at the press conference. The press conference is in fact a summary version of the complete conference with eight short 5 minute presentations, starting with an overview by Hans Meuer explaining, amongst many other things, the difference between this relatively "small" event in comparision to the US Supercomputing conference counter part. This European conference is more focused: no parallel sessions, limited exhibition, so every attendee can meet every other attendee. Then Thomas Sterling previewed his first "looking back" presentation. This should become a "tradition" in Heidelberg. Interesting view: Cray climbs to second place in the capability computing market. Summary of the past year: Infiniband, Itanium, Bluegene, AMD Opteron, Grid computing in Europe, Chinese Dawning, MPICH-2, OpenMPI with fault tolerance, DARPA Petaflop/s, DARPA ACIP, Windows for HPC. That is all. So no dramatic advantages, but constructive continuity Sterling concludes. On the TOP500, Erich Stohmeier reported earlier today. In 2015 we will have a 100 Petaflop/s machine. (1 Petaflop/s = 1000 * 1 Teraglop/s = 1000 * 1 Gflop/s). (AE)

Advertisement

Advertisement

As a main sponsor, AMD's Francesco Toricelli summerized the efforts they did to support the event. One will be a party at Heidelberg's famous castle tomorrow. Sorry if you cannot be there. (Although through the nearby Frankfurt airport it should be possible to get to Heidelberg from most parts of the world within one day.) Another more intersting one is the ISC award.

Frank Baetke did not include HP slides in the Press Kit: on purpose so we have to stay alert. HP SFS is a new HP announcement. A scalable file system based on Lustre. New customers in Europe include HP2CN in Sweden: 384 Opteron processors. SSCK in KArlsruhe: heterogenous cluster with storage works. SPACI - Grid effort in Italy. It is not yet in EnterTheGrid. I will have to look it up and include it there.

In the Hot Seat session tomorrow, there will be presentations by the other companies. Let us wait for that.

Steve Wallach was doing a short version of his presentation from two hours ago. Wolfgang Becker doing a preview of his presentation about SAP and supercomputing. Seems like everyone is wondering what they have to do with each other. But, exactly ten years ago, in 1994, Rainer Zinow at the supercomputing conference which was then still in Mannheim, the chairman of SAP was explaining why the IBM SP2, the first succesful IBM parallel cluster computer was perfect for SAP R/3.

Questions? Yes, I asked about Zinow, of course. The answer from the SAP representative was that today the advances in software, networking, prosessors, are such that now SAP on supers is possible. I will mark this for 2014.

Steve Wallach also pointed out there is now much more off-the-shelf everything and most of all faster networking. Technology has been around for quite some time in the research labs, but only recently came out.

Advertisement

Ad Emmen

EnterTheGrid - Primeur

James Stewartstraat 248

1325 JN Almere

The Netherlands

http://EnterTheGrid.com

mailto:primeur@hoise.com