|
|
||
| 09 June 2003 |
EuroFlash no. 560
|
|
|
Advertisement
Primeur-EnterTheGrid is the premier news service on HPCN, supercomputing and The Grid in Europe. Primeur Weekly delivers the news each week in your e-mail box. Check out http://EnterTheGrid.com for all information on The Grid. Check out the Primeur HPCN web site for the Calendar, the Analysis section with background on the TOP500, the Monthly en Live! special issues, information on HPCN centres and industry.
Advertisement
|
||
Advertisement
The Rutherford Appleton Laboratory (RAL) in Oxon, UK, has completed the upgrade of a 16-node, 32-processor high-performance computing (HPC) cluster christened "Hrothgar". The high-speed SCI (scalable coherent interface) interconnect supplied by Dolphin Interconnect, combined with AMD Athlon processors, is the key to the 71 Gflops performance achieved by the cluster. RAL is the largest laboratory of the Council for the Central Laboratory of the Research Councils (CCLRC), a provider of large-scale facilities and expertise in support of research throughout the UK.
The Large Scale Biosphere-Atmosphere Experiment in Amazonia has been running for the last five years, trying to answer questions on whether it acts as a carbon sink or a source and what are the effects of the recent changes of land use (deforestation) on the rest of the world's environment and consequently on global warming. Below, Chris Lazou interviews Dr. Carlos A. Nobre, the scientific co-ordinator of the Amazonia project, who would like a supercomputer similar in power to the Earth Simulator to analyse their results. (Chris Lazou)
The HealthGrid Association organises the 2nd European HealthGrid Conference at Clermont-Ferrand, France, on January 2004, 29-30th. Deadline for paper abstracts is September 30, 2003.
The University of Liverpool has unveiled its new large-scale Dell high-performance computing cluster (HPCC), powered by Intel Pentium 4 processors, to underpin further ground-breaking disease mapping and physics research, and to enhance international public-interest work conducted on the University's own cross-campus Grid network (ULGRID).
Fujitsu Systems Europe has selected Sistina GFS as the exclusive cluster file system running on Fujitsu's line of High Performance Computing (HPC) clustering solutions. Through this partnership, Sistina and Fujitsu will provide customers and partners with Fujitsu's first pre-configured and packaged Linux cluster solution for both industrial computing applications and enterprise-class data centre deployments.
The PharmaGRID Retreat will be organised by the Prism Forum June 30th-July 2nd, 2003 at the Node Conference Centre, Welwyn, United Kingdom. This workshop brings together industry professionals, stakeholders and GRID practitioners to develop a shared understanding of opportunity represented by the GRID to the pharmaceutical industry.
The Fifth European Workshop on OpenMP (EWOMP'03) will be held September 22-26, 2003 in Aachen, Germany, hosted by the Center for Computing and Communication of the Aachen University. This series of workshops is designed as a forum for discussion of the latest developments in OpenMP, in its compilers, tools and applications. It is part of the cOMPunity initiative whose main objective is the dissemination and exchange of information about OpenMP.
You will not find the hpcLine in the upper regions of the TOP500, but for over four years, this "made in Europe" parallel machines is finding its way in research institutes and industry - mainly engineering - as a reliable workhorse for technical and scientific applications. As Eric Schnepf from Fujitsu Siemens, the manufacturer of the hpcLine, explained at the SAVE meeting at CEA Saclay, 22nd May 2003, the past year, a number of new organisations have started to use the hpcLine, including several in the process and automotive industry. In total there are 115 hpcLine installations with some 3600 processors. Several new Intel and AMD processors were introduced to the machine. Currently new technology, as Infiniband and the upcoming Itanium processors are under evaluation at test customer sites.
Fujitsu has a long history of supporting technical and scientific applications. In the Technical Computing Facility in Paris, the main applications in the field are benchmarked together with the Independent Software Provider (ISV) that develops the code. Goal is to understand the interface between application and machine and see how one or the other can be optimized. In the past year, many of the codes have been tested on the hpcLine. Fujitsu now also offers the applications on the hpcLine a VPP5000 and a Primepower server as Application Service Provider (ASP). This way one can test and even develop ones own applications using state-of-the art hard- and software.
At the SAVE meeting in Saclay, France, in May, Mr. Keller gave an overview of the different experiences with Itanium2, Opteron and Infiniband at the Paderborn Center for Parallel Computing. Paderborn has three systems in use: a A Xeon cluster, purchased from Dell in 2003; a Hewlett-Packard Itanium cluster dating from 2002; and an Opteron cluster, acquired from AMD in 2003. The benchmarks were performed by Dr. Jens Simon in Paderborn.
For HPC computing, one needs 64-bits. Although some scientific and technical work can be done in 32-bits, this is not enough for real large-scale applications. This is one of the reasons, the traditional RISC processors are still dominating that landscape. Intel and AMD now both have processors to handle 64-bits. However, the approach of the companies to implement that in chip design is very different. At the SAVE meeting in Saclay, France, in May, both companies were present, highlighting their new products and developments. And the winner is...
The French energy research centre CEA has a 48-processor hpcLine installed for a CFD project for nuclear applications. At the SAVE meeting in Saclay in May, Christophe Calvin from CEA talked about the experience they had with the machine.
During the past year, Scali has redefined and extended its product suite, focusing on taking the difficulty out of optimising a parallel application onto different kinds of interconnect hardware. The products are now called "Scali Manage" - Cluster Management Solution and Scali MPI Connect - High performance Inter-process communication. The Oslo based company also acquired a second tranche of financing earlier this year.
Axel Keller described the PadFEM2 approach used by Stephan Blazy at the Paderborn Parallel Computing Center in a project on massive parallelising for the balancing of dynamic networks, led by Professor B. Monien, at the SAVE meeting in Saclay, France, in May. For the application of PadFEM or parallel adaptive finite element methods, the Fujitsu Siemens hpcLine and an Itanium2 based system are used as cluster, Grid and numerical computing resources.
A AU$1.7 million, supercomputer called Hydra will be unveiled at the University of Adelaide today by the Science and Information Economy Minister, Dr. Jane Lomax-Smith. With a peak of 1.2 Tflop/s, Hydra is among Australia's top three supercomputers and is one of the fastest of its kind in the world, offering unique benefits to South Australian researchers and industry.
The Eighth Global Grid Forum (GGF8) is being held June 24-27 in Seattle, Washington, USA. The advance registration deadline is Friday, June 13th. By registering in advance, attendees can add tutorials at ~50% discount and attend the HPDC-12 immediately before GGF8 participation at a reduced fee. HPDC-12 will be held in Seattle as well. Here, the deadline is also Friday, June 13.
The US National Weather Service (NOAA) has activated its weather-forecasting supercomputer, a massive machine with IBM eServer and storage systems that will extend the government's ability to forecast hurricanes from three days to five days in advance. Making its debut as hurricane season 2003 begins, the machine is expected to give businesses, government agencies and citizens crucial extra time to prepare for the violent storms.
The Texas Advanced Computing Center (TACC), a research facility at the University of Texas at Austin, has selected Topspin's Switched Computing System as a planned key component of its high-performance visualisation cluster.
ANSYS has released CFX-5.6 which provides increased accessibility to advanced computational fluid dynamics (CFD).
Sun Microsystems says regained its position as the worldwide leader in Linux+Unix server revenue and unit shipments in the first quarter of CY2003, according to the 1Q03 Worldwide Server Database by Gartner Dataquest.
The Sixth International Workshop on Performability Modelling of Computer and Communication Systems (PMCCS-6) will be held September 5-7, 2003 in Monticello, Illinois, USA. The workshop is part of the 2003 Illinois International Multiconference on Measurement, Modelling, and Evaluation of Computer-Communication Systems. The deadline for paper abstract submission has been extended to June 16, 2003.
Dot Hill is partnering with system integrator ASL Automated Ltd in Thailand. ASL, based in Bangkok, provides technology solutions and services to its extensive customer base. ASL also works closely with specialised vertical solution providers in the financial, insurance and health care sectors.
Primeur Weekly is published per e-mail. Check out the subscription information for more details how to subscribe. You can find the back issues on the Primeur web site. The EuroFlash! and USflash! are published together with ESIS - European Supercomputing Information Service.
© 2003, Genias Benelux