| News digest 20 June 2008 |
|
|
 |
| |
| Start |
|
|
| Primeur Live! from Dresden |
|
Continue over a decade of PrimeurLive! reporting from Eruoep's major supercomputing event, Primeur publishes this three PrimeurLive! issues from ISC08 in Dresden:
|
|
|
|
Advertisement
|
|
| |
| Blog |
|
|
| Cray's update |
|
Steve Scott from Cray did provide a quick overview of Cray activities and plans. Today Europe represents about 30% of the Cray market, and for instance the Cray 2008 user group meeting is in Helsinki. In the US they are working on a USD 250 million DARPA HPCS contract and they did win a significant part of the TI-08 , NSF Track-2 system at UT. The Cray supercomputer at NERSC is now running at 100 Tflop/s The Cray at ORNL the system will be upgraded to Pflop/s performance later this year. A new partnership between Intel and Cray has been announced. It is a collaborative agreement to pursue high-end HPC technologies and opportunities. But Steve Scott said this will not influence the partnership with AMD.
|
|
|
|
| Supercomputers or Grid? |
|
The performance was not exactly a "Battle-Rap", but it came very close. Wolfgang Gentzsch and Dieter Kranzmueller gave at the closing day in Dresden "battle" about Grid computing versus Supercomputing. Everybody knows Wolfgang Gentzsch as mr Grid: Genias, Gridware, SUN Grid engine, D-Grid. But he changed camps and is now dissemination manager of DEISA the European supercomputer collaboration project. Dieter Kranzmueller was active in supercomputing for some time, but today is involved in the European Grid Initiative. Wolfgang kicked of with "Grids are dead" but Dieter replied immediately with recent Grid successes, such as calculations for the Avian Flu, that were set-up very fast and did lead to useful results, and WISDOM, that helps to contribute to solving Malaria.
Read further... |
|
|
|
| Fujitsu - we wil be back sometime in the future |
|
In the Hot seat session Fujitsu mainly concentrated on the recently introduced FX1 supercomputer. the first system was installed at the Japan Aerospace Exploration Agency (JAXA). The road map of the FX1 shows a 100 Tflop/s machine. Fujitsu is also participation in the Japanese effort led by RIKEN to produce a 10 Pflop/s system in 2010. The company is building the scalar processor for that machine. But before that they hope to see a number of FX1 systems in the TOP50 next year.
|
|
|
|
| NEC to introduce Assignable Data Buffer |
|
Dr. Rudolf Fischer from NEC, Germany explained where we are in HPC and how we have to move forward. Last October, the NEC SX-9 was launched. This machine delivers 102.4 GFlops with 8 results per cycle. An innovation is the Assignable Data Buffer (ADB) but it is not a cache but makes use of data locating. It is shared by the Scalar unit and Vector unit. It is the most powerful UMA SMP node in the world.
Read further... |
|
|
|
| The success of SGI in 2007 |
|
Bill Mannel from SGI recalled the successes of the past year for SGI. The SGI Altix ICE 8200 was announced June 26th, 2007, which meant a new era in HPC computing with an innovative approach. The machine integrated the best of both worlds with up to 512 cores and 6 TFlops per rack. SGI also optimised the cooling and power efficiency. The company introduced the water-cooled coils. SGI Tempo is an innovative cluster management.
Read further... |
|
|
|
Advertisement
|
|
| |
| TOP500 |
|
|
| Argonne's supercomputer named world's fastest for open science and third overall |
|
The United States Department of Energy's (DOE) Argonne National Laboratory's IBM Blue Gene/P high-performance computing system is now the fastest supercomputer in the world for open science, according to the semiannual Top500 List of the world's fastest computers. The Top500 List was announced today during the International Supercomputing Conference in Dresden, Germany.
Read further... |
|
|
|
Advertisement
|
|
| |
| Hardware |
|
|
| Challenges of the Petaflops era |
|
Thomas Sterling talked about HPC opportunities in the PFlops era. The greatest challenge is to bring Petaflops to real world applications and problems. The exploitation of ubiquitous arithmetic and logic units now happens with multi- and many-core technologies, accelerators and GPU. There are new packaging techniques for higher density such as blades and 3D die stacking. The performance to cost advantage is also important with commodity clusters, attached accelerators, and so on.
Read further... |
|
|
|
| |
| The Grid |
|
|
| Action needed for a grown up HPC infrastructure in Europe |
|
Dr. Kimmo Koski from CSC, the Finnish IT Center for Science, organized a Bird-of-a-Feather session on setting up an HPC Ecosystem in Europe, a computational science infrastructure that will meet the supranational needs within the scientific community in the years to come. Rob Pennington from the National Center for Supercomputing at the University of Illinois at Urbana-Champaign, described infrastructure as a global challenge taking into account the design complexity, software, cost and power cooling, and the interoperability across administrative units.
Read further... |
|
|
|
| We need to build our own supercomputers |
|
Dr. John Salmon from D.E. Shaw Research stated that his company feels the need to build its own supercomputers. In the next couple of days it will celebrate its 20th anniversary. Founder of the company first worked in Wall Street. D.E. Research is an independent research laboratory in computational biochemistry. There are two approaches, namely laboratory experiments and biophysical simulation.
Read further... |
|
|
|
| |
| Company news |
|
|
| Fujitsu supercomputers make TOP500 list |
|
Thanks to their outstanding execution efficiency, supercomputer systems constructed by Fujitsu Limited were included in the latest edition of the TOP500 list of the world's fastest supercomputers, announced on June 18 at the International Supercomputing Conference (ISC'08) in Dresden, Germany. The inclusion reflects the high-performance software development technologies offered by Fujitsu combined with superior systems integration technology that extracts the maximum performance from the hardware.
Read further... |
|
|
|
| ADVA Optical Networking, Obsidian and Voltaire prove support for Infiniband over 50km |
|
ADVA Optical Networking, Obsidian Strategics and Voltaire have achieved an integrated capability to extend InfiniBand connections to 50km without performance degradation. The solution was successfully tested at the University of Stuttgart's High Performance Computing Center (HLRS), which provides systems, tools and expertise for public and private supercomputing development projects.
Read further... |
|
|
|
| EMCORE announces sampling and demo of 40 gigabit per second connects cables for high-performance computing |
|
EMCORE Corporation, a provider of compound semiconductor-based components and subsystems for the broadband, fiber optic, satellite and terrestrial solar power markets, has performed a sampling and demo of 40 gigabit per second (Gb/s) fiber-optic transceiver-based cables that are being showcased with Mellanox's switches and adapters at the International Supercomputing Conference in Dresden.
Read further... |
|
|
|
| IBM sets the pace for TOP500 |
|
IBM showing occupying in the TOP500 the top three spots and a total of 210 systems on the list - the most of any supercomputer vendor. IBM also had the most aggregate performance on the list with 5.6 petaflops (48% of total); and the most systems in the top 10, top 50, and top 100.
Read further... |
|
|
|
| Voltaire and DataDirect Networks deliver native InfiniBand storage to customers worldwide |
|
Voltaire, a provider of Grid backbone solutions for data centres, is working with DataDirect Networks Inc., a data infrastructure provider for the most extreme, content-intensive environments in the world, to deliver native InfiniBand storage to customers worldwide. The companies recently completed cross-certification testing of Voltaire Grid Director InfiniBand-based switches and software with DataDirect Networks S2A9900 StorageScaler systems. Used together, the products deliver up to a 200 percent improvement in storage performance compared to solutions that use Fibre Channel.
Read further... |
|
|
|
| Voltaire switches to support new Engenio 7900 HPC storage system from LSI |
|
Voltaire will provide support for the next-generation Engenio 7900 HPC storage system introduced by LSI Corporation. The Engenio 7900 HPC enables customers to use Voltaire's InfiniBand-based switches and software to directly attach to storage for a high-performance, low cost storage solution.
Read further... |
|
|
|
| Sun expands Sun Constellation System Family |
|
New Sun Constellation System additions enable customers to scale from a single rack at approximately 7 TFlops to over 2 PFlops within the same, compatible architecture - allowing customers to deploy a scalable high-performance cluster in any data centre. Sun also announced a new Storage and Archive solution for HPC and updated software and development tools that simplify HPC cluster installation and management.
Read further... |
|
|
|
| LSI Storage System brings enterprise-class availability and reliability to HPC environments |
|
LSI Corporation has introduced the Engenio 7900 HPC storage system which combines leading-edge performance and high levels of system availability for applications requiring uninterrupted data access, designed for compute-intensive applications and high bandwidth workloads.
Read further... |
|
|
|
| Acceleware announces GPU based cluster solution |
|
At the International Supercomputing Conference, Acceleware Corp., a developer of high performance computing (HPC) solutions, has released the world's first commercially available GPU-based cluster solution, the C30-16. This solution combines Acceleware's new clustering technology with its portfolio of designed-for-parallel computational algorithms to harness the power of 64 GPUs, delivering unprecedented performance and scalability for enterprise customers.
Read further... |
|
|
