The year 2002 in Grid and HPC: from Computenik, hypes and Kites

Almere 31 December 2003 The Earth Simulator rocketed into the first position of the TOP500 in June. Cluster computers invaded the TOP10 in November. That were the two main news items in supercomputing in 2002. In Grid computing everyone continued relabling meta-computing, distributed computing, p2p computing and the like into "Grid computing". The waiting for and working on standards was the main development. All major countries like for instance the UK, France and Italy now have national Grid programmes under way. With a record of 35 Gbit/s into one location in Amsterdam during iGrid2002 and a lot of applications running on it, it looks like indeed the network is not the bootleneck for high-end Grid computing anymore: it is the application. (Flying kites that where calculated on different locations over the world and displayed in a Virtual Reality Cave in Amsterdam, showed it is not only eScience and commercial Grid computing that use grid resources, it can also be applied for arts.) Of course, most of use, especially in industrial Grid computing, do not have that type of band width yet. A selection of the articles illustrating the main developments that were published in Primeur/EnterTheGrid in 2002.

	Japanese Earth Simulator takes first position in TOP500
	Japanese 'Computenik' Earth Simulator shatters US supercomputer hegemony
	The Japanese Earth Simulator is on-line and producing results that alarm the USA, that considered itself as being leading in supercomputing technology. With over 35 Tflop/s, it five times outperforms the Asci White supercomputer that is leading the current TOP500 list. No doubt that position is for the Earth Simulator, not only for the next list, but probably even for the coming two years. In the New York Times, bench mark compiler Jack Dongarra compares the event with the Sputnik, hence he dubbed the Earth Simulator "Computenik".
	Full article (May 2002)...
	Blue Planet to challenge Earth Simulator Computenik
	In April this year, the Japanese Earth Simulator rocketed into the position of fastest supercomputer in the world. The difference with the number 2 in the TOP500 list is so big - even the combined power of the first dozen of US made supercomputers together did not match the Earth Simulator, that the machine was dubbed Computenik, shattering the US supercomputing hegemony. Since this event, US scientists and government are looking for an answer. One of the more interesting proposals is the "Blue Planet". In a proposal from Lawrence Berkeley National Laboratory, Argonne national laboratories, and IBM, Blue Planet is developed as a machine that could deliver twice the Earth Simulator performance at half the cost in 2005. This should bring a US machine back on top of the lists. If the Japanese do not do anything in the mean time that is.
	Full article (November 2002)...
	Clusters enter TOP10 of November 2002 TOP500 list
	Scientists from Heidelberg have built HELICS, one of the worldwide fastest parallel PC clusters
	The Interdisciplinary Centre for Scientific Computing of the University of Heidelberg (IWR) has just finished the installation of a parallel high speed PC cluster with an excellent, yet unknown price/performance ratio. This parallel high-performance computer, named HELICS - the Heidelberg Linux Cluster System - which has been acquired together with the computing centres of the University of Mannheim (RUM) and the University of Heidelberg (URZ), is set up with standard components, so-called "Commodity off the Shelf" components. It consists of 512 AMD Athlon MP PC processors; two of them are placed into one computing node. These processors have frequencies of 1.4 GHz and reach a theoretical peak performance of 2.4 billion floating point operations per second (Gflops) for each processor.
	Full article (May 2002)...
	20th Edition of TOP500 List of World's Fastest Supercomputers Released
	In what has become a much-anticipated event in the world of high-performance computing, the 20th edition of the "TOP500" list of the world's fastest supercomputers was released today (November 15, 2002). The Earth Simulator supercomputer installed earlier this year at the Earth Simulator Center in Yokohama, Japan, is with its Linpack benchmark performance of 35.86 Tflop/s (trillions of calculations per second) retains the number one position. The #2 and #3 positions are held by two new, identical ASCI Q systems at Los Alamos National Laboratory (7.73 Tflop/s each). These systems are built by Hewlett-Packard and based on the AlphaServer SC computer system.
	Full article (December 2001)...
	Number of Tflop/s systems in Europe tripled in six months
	The new November 2002 TOP500 list of the world's most powerful supercomputers released shows that there are now 176 TOP500 class supercomputers in Europe, compared to 171 six months ago. The HP/Compaq Alpha server at CEA in France is still the fastest machine in Europe, and now on position 7 world-wide. There are 12 systems in Europe with a computing power of over 1 Tflop/s. Six months ago, there were only 4.
	Full article... (December 2002)
	Linux NetworX supercomputer ranked as fifth fastest in the World
	MCR, the cluster supercomputer Linux NetworX built for Lawrence Livermore National Laboratory, has been ranked as the fifth fastest supercomputer in the world on the TOP500 supercomputing list. The 2304-processor cluster can process 5.7 Tflop/s running the Linpak benchmark, and is the only Linux-based supercomputer to be ranked within the top five. A 361-node cluster Linux NetworX built for Argonne National Laboratory also broke the Tflop/s barrier, capable of 1.6 Tflop/s and is ranked 46th on the TOP500 supercomputing list. The TOP500 lists the 500 most powerful supercomputers in the world based on Linpak benchmark results.
	Full article (December 2002) ...
	Quadrics Top in Top 500 and IDC Rating
	Analysing the Top500 under the perspective of high-end clusters, Quadrics ranks with 3 of the top 5 as the network of choice in high-end computing. These machines use Quadrics QsNet as the system area data network. The same is true with the IDC Balanced Rating Benchmark of Capability Computers from November 2002. Here 5 of the leading 15 systems use Quadrics QsNet, although IDC deleted the second machine of Los Alamos National Lab, as both are identical. (Uwe Harms)
	Full article (January 2003)...
	European contributions to the Grid
	Overwhelming participation to Global Grid Forum in Edinburgh feeds fear for overhyping
	The biggest threat to Grid is overhyping. That is the fear of Global Grid Forum (GGF) president Charlie Catlett. At the fifth Global Grid Forum (GGF5) in Edinburgh, Scotland, he welcomed close to a thousand participants. Far more than expected. The standardisation process going on in GGF is, by the very nature of it, slow. The organisation of GGF is in place, said Catlett, the first documents will enter the formal standardisation process soon, but we cannot rush that process. Good interoperable standards are key for creating a healthy Grid market with several competing but interoperable products. This opinion was also advocated by a number of other presenters in Edinburgh.
	Full article (August 2002) ...
	From Supercomputers to PDAs: Cactus on an iPAQ
	Cactus was originally designed as a framework for physicists with largescale applications which, to model complex natural phenomena with adequate fidelity, need the kind of computational resources only supplied by the world's largest supercomputers. One important design requirement has been to ensure that the physicists developing and running their simulations had a common working environment. Their codes should be able to be developed and debugged on local laptops and workstations, and then be immediately run on any final production machine, whether it be a Linux cluster, a Cray T3E or any other supercomputer.
	Full article (October 2002)...
	The new extreme-bandwidth standard: 35 Gbit/s into one location during iGrid2002 in Amsterdam
	Never before, so much bandwidth has been available to one place on earth as to the WTCW centre in Amsterdam during iGrid2002 last week. No less than 35 Gbit/s bandwidth was available for connections to the USA and Geneva: that is a million times more than an ordinary dial-up connection offers. Even the current plans for SC2002, the world's largest supercomputer exhibition, that will be held later this year, do not yet foresee this type of bandwidth. This type of bandwidth was made possible by using new generation Lambda routers. There was so much bandwidth available that the applications demonstrations had a hard time to fill it up. This despite the fact that there were a number of very bandwidth-intensive applications, like multi-Cave interactions and multi-channel HDTV over Internet. Playing music together with people in Chicago, flying kites that are calculated around the globe, dancing with people in other Caves were some of the non-scientific applications.
	Full article (October 2002)...
	Kites Flying In and Out of Space exhibited at iGrid2002
	Jackie Matisse's "Kites Flying In and Out of Space" is the first high bandwidth art piece ever created. Exhibited at the iGRID2002 Conference hosted by SARA in Amsterdam, The Netherlands, September 23-26, 2002, "Kites Flying In and Out of Space" utilises a "Grid" model for real time steering of calculations on computers distributed over high-speed networks. Each of the 12 kites appearing in the piece utilises up to 15 megabits per second. This art piece uses a total of approximately 180 megabits per second in calculating the forms and theoretically could utilise even more. CAVEs around the world could potentially view this application through a connection to the Starlight high-speed networking programme.
	Full article (October 2002)...
	Unicore is an "OGSA safe" integrated Grid platform
	The German Unicore plus project has delivered the Unicore 4.0 vertically integrated Grid environment. Unicore 4.0 is an industry-strength platform that has been developed during the past five years with support from the German government. At the final meeting in November 2002 in Bonn, David Snelling, one of the architects of Unicore, showed a road map on how to get Unicore "OGSA compliant". Althought the Open Grid Services Architecture (OGSA) is a standard under development, David Snelling showed that the basic structure of OGSA and Unicore are very similar. This is remarkable considering that at the time when Unicore was conceived, Grid computing was not in focus. The Unicore software is not as widespread yet as the Globus toolkit. Main reason, in my opinion, is Globus comes from the USA and Unicore does not. Also there is the misconception of Unicore not being "open source". In a number European Commission funded projects, like EuroGrid, Grip and OpenMolGrid, Unicore is used, however, as in the UK project RealityGrid. Apart from the stability and robusteness, the feature that allows to create an application plug-in for Unicore which allows to present the user his usual application environment with Unicore functionality proves useful.
	Full article (January 2003)...
	The Grid is a fractal - Sophisticated brokerage is needed to glue it together
	"The Grid" as such does not exist. There are many small Grids, which sometimes can co-operate, but most of them are not even aware of each other. Hence "The Grid" can be considered as a worldwide conglomerate of small fractions that have dynamics of their own. Eventually, many of them will be connected to each other. One of the tools that are needed for this to happen is a universal brokerage mechanism. A Grid broker can take a request from a user to run, for instance, a computer job, get quotes from Grids and show the user the possibilities. At the Unicore Symposium in Bonn Germany, John Brooke, from the University of Manchester, shared his ideas on the fractal nature of the Grid, the need for brokerage, and the resource broker system that is being developed at the University of Manchester by Jon MacLaren and which is used in, for instance, EuroGrid.
	Full article (January 2003)...
	European countries go Grid
	French government heavily invests in powerful computing and high-speed networks for competitive research
	The French Ministry of Research has turned the access for researchers to larger computer systems and information and communications technologies into a first-class priority. Whatever the scientific disciplines may be, ranging from physics to medicine or social sciences, the use of powerful information tools has become indispensable to the knowledge progress and to its effects on society, pleaded Catherine Roucairol, representative of the French Ministry of Research, at the Fourth DataGrid Conference in Paris. The understanding of large climatological or economic phenomena as well as the analysis of the human genome and matter in general require the increasing use of the most powerful computers in the world and the largest data storage capacities. High-performance tools combined with the use of high-speed networks constitute the necessary equipment for scientific research. International competitivity and the efficiency of the French Research are directly depending on it.
	Full article (February 2002)...
	The UK e-Science programme
	In the UK, rather than concentrating on The Grid itself, the broader focus of e-Science has been chosen. The UK e-Science programme develops support for large-scale science through distributed global collaborations. Within thta goals, One focus of the e-science programme is the development of a Grid communication and computational infrastructure to underpin the work of scientists. In the UK vision, the Grid promotes the rapid formation of virtual colaboratories allowing scientists to work together and share resources irrespective of the location of the scientists or the resources they are using.
	Full article (May 2002)...
	Italy wants BIGEST Grid
	In Italy, there are several initiatives to create large scale Grids. The first is a recently accepted proposal called MIUR-FIRB that is co-ordinated by the Italian National Research Council (CNR). MIUR-FIRB will enable ICT platforms for distributed high-performance Computational Grids oriented to the development of complex multi-disciplinary applications and scalable Virtual Organizations. A second initiative, co-ordinated by the Italian high-energy physics organisation INFN, called Italian Grid for Business, Industry, Government and E-Science and Technology (IG-BIGEST), aims at the creation of an Italian Research Area (IRA) and supporting infrastructure. This is an Italian e-Science initiative, currently in proposal stage.
	Full article (September 2002)...
	UK Government announces a further GBP 31.6 million for PPARC's e-Science Programme
	Following the increased funding for science announced by the government today, the Particle Physics and Astronomy Research Council (PPARC) has received a major uplift of GBP 25 million in its baseline budget. In addition PPARC will receive continued investment in its e-science programme, and specific allocations to increase its investment in accelerator R and D, gravitational waves, and planetary exploration. PPARC will also receive funding to implement the recommendations of the Roberts report and to upgrade the infrastructure of its institutes.
	Full article (January 2002)...
	Primeur/EnterTheGrid highlights
	EnterTheGrid adds Knowledge Management to world's largest Grid catalogue
	EnterTheGrid, the largest catalogue on Grid computing in the world has renewed its web site and added knowledge management tools that help people identify Grid products. The new knowledge management tools generate listings from the Grid related information taken from different viewpoints and providing insight into the relationships between them.
	Full article (August 2002)...
	Information from TOP500, EnterTheGrid used to enrich supercomputer overview report
	Recently, the report "Overview of recent Supercomputers - 2002" by Aad van der Steen has been published. It provides a detailed description of supercomputer systems. In this way, it complements the TOP500 list, that tells which machines or the most powerful currently and EnterTheGrid, that provides a description of the companies. However, if one is interested in combining the information from these three sources in their original formats, that is a tedious job. However, we made all three sources available in XML-format. Combining information and extracting knowledge is then an easier task. We did produce a version of the report that, for each supercomputer architecture section in the Overview report, also lists the TOP500 entries in the June 2002 list, and the latest company description from EnterTheGrid. Summary information has been extracted from the XML-sources and is presented too. The thus enriched report is available as a PDF-file.
	Full article (September 2002)...
	EnterTheGrid - a business analysis tool for the Grid
	Grid computing is a fast growing dynamic area. Depending on which analyst you talk to, the Grid will grow in double or even triple numbers in the coming years. Having the right business and marketing information at hand, can help you enter or stay in the Grid market. EnterTheGrid provides you with the marketing and business information you need. EnterTheGrid offers access to the largest Grid catalogue in the world on a yearly subscription base.
	Full article (October 2002)...
	More than 50,000 visitors to server that runs Primeur/EnterTheGrid
	For the first time, the HOISe server that runs Primeur/EnterTheGrid, received more than 50,000 unique visitors in one month. This can be concluded from the November 2002 server statistics. This makes Primeur/EnterTheGrid the largest news source that reports on both HPC and Grid computing in the world. Visitors come from all over the world, including Europe, the Americas, and Asia. Primeur/EnterTheGrid offers a weekly news magazine for subscribers, a free monthly news digest, and EnterTheGrid.com, the largest catalogue on Grid Computing in the world.
	Full article (January 2003)...

[News on Advanced IT]   [Calendar]   [Analysis]   [IT in Medicine]