logo

EnterTheGrid - Primeur Live!

EnterTheGrid - Primeur is the premier Grid and Supercomputing information source in the world.

>Primeur Magazine
>PrimeurLive!
>EnterTheGrid
>Analysis
>Backissues
>Calendar
>Subscribe
>Advertise
>Contact
News digest 22 June 2005
>Start
>PrimeurLive! from ISC2005 in Heidelberg
>Blog
>Sad and happy days
>Why a TOP500, why not TOP100 or TOP1000 supercomputers
>From the exhibition floor
>TOP500
>25th Edition of TOP500 List of World’s Fastest Supercomputers Released
>Twenty year anniversary of supercomputer history in market statistics
>MareNostrum, the building of an icon in a temple
>Columbia Supercluster at NASA has already 700 users
>Hardware
>High density computing and enriched programming methods major current trends in high performance computing
>Ten factors causing dramatic change in 20 years of supercomputing and future challenges
News digest 22 June 2005
Advertisement
Visit our sponsors
Start
PrimeurLive! from ISC2005 in Heidelberg

This year we again report Live! from Europe's main supercomputing event. See the issues for each day:

As last year we start each issue with an "blog" giving you some personal impressions. The next sections provide traditional reporting. The last section "Company news" provides mainly press release type of information from the companies. Hence, a few items are brought to you from three different view points.
Advertisement
Visit our sponsors
Blog
Sad and happy days
The International Supercomputer Conference is at its start, already more successful than ever, Hans Meuer said in his opening speech of ISC2005 in Heidelberg. At the 20th anniversary meeting about 625 particants have been registered. About 70% from the particapants are coming from Europe the rest from all over the world, including for the first time people from emerging supercomputer nation China. Hans Meuer showed that the conference and accompanying exhibition grew steadily over the past 20 years. Apart from this good news, Hans Meuer also did acknowledge some sad events. Hans-Martin Wacker, cofounder of the Supercomputer conference, died in July last year: he would have loved to be at the20th anniversary Supercomputing conference. A good friend and a well-known person in the community, Hans said. The Conference book contains an Obituary in rememberance. Another person who was already present at the first conference twenty years ago was Uwe Harms, who died ten days ago in his home in Munich. Hans recalls he was the unofficial public relations office of the whole HPC community. Just five years ago, he presented an overview of the first 15 supercomputer conferences. His last article was published in the latest issue of CADplus magazine about high-performance computing and the upcoming conference.
Why a TOP500, why not TOP100 or TOP1000 supercomputers
Ever wondered why the TOP500 is a TOP500 and not a TOP1000 or a TOP100. Well, we knew at the time when we started the TOP500, there were at least some 500 supercomputers in the world, Erich Strohmaier said during his traditional overview of the TOP500. And, seeing the trends and statistics that still seems a right choice today. There are neverending discussions about the TOP500, especially by vendors who do not appear at the top in the latest lists, about how well the simple single number of the Linpack benchmark is a correct representation of a specific machine in the list. In the statistics of the complete list of 500 machines, the anomalies smooth out. Erich Strohmaier showed that over the years, the TOP500 has been good at consistently capturing the trends in for instance architectural changes, trends for large countries, continents, etc. There are things you cannot do with the TOP500. For instance because of the fixed size of 500 machines, it cannot capture market size developments.
From the exhibition floor
The exhibition floor in Heidelberg is crowded. In fact they could have easily sold exhibition space twice as large. That is one of the reasons, along with the fact that the world soccer games are in Germany in June next year, that the ISC conference will move to Dresden, in the eastern part of Germany. The large conference centre can accomodate a larger exhibition. Today, we visited and talked to Intel and Pathscale. Both are companies that deliver HPC hardware and software components, but this is where the similarity ends. Intel showed a first 4-processor Montecito based platform with 46 Gflop/s performance, a world premiere. Pathscale exhibits its new low latency InifiniPath Infiniband connector. Read further...
Advertisement
Visit our sponsors
TOP500
25th Edition of TOP500 List of World’s Fastest Supercomputers Released
In what has become a closely watched event in the world of high-performance computing, the 25th edition of the TOP500 list of the world’s fastest supercomputers was released today (June 22, 2005) at the 20th International Supercomputing Conference (ISC2005) in Heidelberg Germany. Read further...
Twenty year anniversary of supercomputer history in market statistics
At the occasion of the 20th anniversary of the International Supercomputer Conference series in Heidelberg, Erich Strohmaier from the Future Technology Group at Lawrence Berkeley National Laboratory in California, presented an overview of the 20-year history in analysis of supercomputer market statistics. From vector machines in the late seventies and eighties over Massive Parallel Systems (MPP) and Symmetrical Multiprocessor Systems (SMP) in the nineties to cluster concepts in the early 2000s, all have had their successes in the TOP500 project which started off in 1993. At present, new hardware architectures and programming paradigms are being explored based on low power components but delivering surpisingly high performance all the same. And the DARPA HPCS Programme even aims at building a PetaFlops computer before the end of this decade. Read further...
MareNostrum, the building of an icon in a temple
In the Wednesday afternoon session "New Architectures, New Applications" Professor Jesús Labarta, Director HPC Research, at the Technical University of Catalonia Supercomputer Center in Barcelona, held a talk on the advanced technical capabilities of the MareNostrum system, recently installed in a stunning multi-arched university building in Barcelona, Spain, which used to be a church - the University is hosted in a former convent. Since November 2004, MareNostrum, an IBM based project, figures in the Top 5 among systems such as the BlueGene/L, Columbia, the Earth Simulator and Thunder in the TOP 500 list. What more reason can there be to invite Professor Labarta at ISC2005? Read further...
Columbia Supercluster at NASA has already 700 users
Dr. Walt Brooks, Chief of the NASA Advanced Supercomputing (NAS) Division at the NASA Ames Research Center in Moffett Field, California, presented the benchmark results and the capability and capacity performance of the newly installed SGI Columbia cluster at NASA before the ISC2005 audience. At present, Columbia has no less than 700 users and it definitely changed the way supercomputing is done at NASA, testified Dr. Brooks. Read further...
Advertisement
Visit our sponsors
Hardware
High density computing and enriched programming methods major current trends in high performance computing
In the "HPC Retrospect" session at the end of the first ISC conference day, Thomas Sterling from the California Institute of Technology, Caltech in Pasadena, highlighted the most important developments in HPC during the last year. Two major examples cited by Mr. Sterling are the evolution towards high density computing and the need for new, more refined models of programming languages. The notion of high density computing should be understood as a major increase in the relative amount of arithmetic functional capability in a supercomputer, according to the speaker. Developers have achieved this with the multicore strategy and also with the ALU arrays approach. As for the programming methods, we now witness the emerging popularity of Co-Array Fortran and UPC, next to the growth of MPI to MPI-2 and this year's birth of MPICH2. Read further...
Ten factors causing dramatic change in 20 years of supercomputing and future challenges
Hans Meuer, ISC2005 organiser, invited keynote speaker Horst Simon, Director at Lawrence Berkeley National Laboratory, to give a refreshing view on the breakthroughs in 20 years of supercomputing and on the challenging problems that need to be solved in the coming decades. Dr. Simon performed the job more than excellent by providing a top 10 of resans why supercomputing dramatically changed between 1985 and 2005 and by describing the challenges faced by the supercomputer community in the time scales covering the periods 2005-2012, 2010-2018 and 2015-2025. Read further...

EnterTheGrid - Primeur

James Stewartstraat 248

1325 JN Almere

The Netherlands

http://EnterTheGrid.com

mailto:primeur@hoise.com

© EnterTheGrid - Primeur Live!