The conference program for the 2009 International Supercomputing Conference (ISC’09)is now finalized, and according to ISC’09 General Chair Prof. Hans Meuer, this year's conference program is set to be the most illustrious in the 24-year history of the conference. This year’s conference will be held June 23-24 in Hamburg.
The four-day high performance computing conference and trade show at the Congress Center Hamburg provides a unique international platform to gain insights, network and do business, all under one roof. With over 1,500 attendees and more than 120 exhibitors from over 45 countries expected to attend, ISC’09 is shaping up to be the most extensive and dynamic since the conference began as a meeting in Mannheim in 1986.
ISC, which marks its 24th anniversary in 2009, has a well-established reputation for presenting well-founded, precise and up-to-date information in an environment that encourages informal conversations and sharing of ideas. ISC is also the largest high performance computing exhibition in Europe. All conference proceedings are conducted in English.
In addition to world-renowned keynote speakers, ISC’09 will feature four in-depth sessions examining some of the most exciting – and challenging – areas in high performance computing today. Here is a quick look at each of the sessions:
HPC & Cloud Computing – Synergy or Competition?, featuring speakers from Google, Yahoo, Amazon, Microsoft and others
Climate Modeling & HPC: Challenges in a Changing World, featuring speakers from leading climate research institutions, including ECMWF, NCAR, MPI, GFDL and COLA.
Supercomputing Challenges for Research in Aeronautics, featuring speakers from EADS, Airbus, Boeing, DLR and Turbomeca.
Scientific Afternoon, with parallel sessions on burning scientific topics, chosen by a panel of judges from submissions by researchers around the world.
“Every year, as we put together the conference program, I say this will be the best ever, but this year the program is truly illustrious,” said Conference Chair Prof. Hans Werner Meuer of the University of Mannheim. “For 2009, we have four days of presentations, an impressive lineup of speakers and our largest exhibition ever. At the same time, we have also updated our pricing structure to ensure that everyone who wants to attend can afford this extremely valuable investment.”
The ISC’09 program also includes four keynote speakers:
Andreas von Bechtolsheim of Arista Networks and the legendary co-founder of Sun Microsystems, will discuss The Evolution of Interconnects for High Performance Computing during the opening session keynote address on Tuesday, June 23.
Prof. Dr. Gunter Dueck, IBM Distinguished Engineer, will give the Thursday, June 25 keynote talk on Lean Brain Management – More Success and Efficiency by Saving Intelligence
An overview of the ISC’09 conference program can be found at http://www.supercomp.de/isc09/Program/Overview.
Online registration starts on March 2
Advance registration for ISC’09 begins Monday, March 2, and in addition to reduced rates, a number of one-day options for both the program and exhibition are being offered for the first time. As an added incentive for U.S. attendees, the U.S. dollar is about 25 percent stronger against the euro than last year. For complete details on registration categories and rates, go to: http://www.supercomp.de/isc09/Registration-Travel/Attendance-Fees
With the passage of a molecule through the labyrinth of a chemical system being so critical to catalysis and other important chemical processes, supercomputer simulations are frequently used to model potential molecule/labyrinth interactions. In the past, such simulations have been expensive and time-consuming to carry out, but now researchers with the Lawrence Berkeley National Laboratory (Berkeley Lab) have developed a new algorithm that should make future simulations easier and faster to supercompute, and yield much more accurate results.
"Currently the major limiting factor in running molecular simulations for a large number of structures before they can be screened for useful materials is the need to visually analyze the structures to set up successful simulations," says Maciej Haranczyk, a computational chemist and a 2008 Glenn T. Seaborg Fellow in Berkeley Lab's Computational Research Division. "With our approach, such structural analysis can be done automatically, which speeds up the whole process of material screening."
Haranczyk is co-author of a paper that appears in the Proceedings of the National Academy of Sciences entitled: "Navigating molecular worms inside chemical labyrinths." The other author of this paper is James Sethian, who heads the Mathematics Group of Berkeley Lab's Computational Research Division, and is also a professor in the Mathematics Department of the University of California, Berkeley.
A key to the success of this new algorithm was its departure from the traditional treatment of molecules as hard spheres with fixed radii. Instead, Haranczyk and Sethian constructed "molecular worms" from blocks connected by flexible links. These molecular worms provide a more realistic depiction of a molecule's geometry, thereby providing a more accurate picture of how that molecule will navigate through a given chemical labyrinth, as Sethian explains.
"In practice, most molecules of interest, even the simplest solvents or gases, rarely have a spherical shape, and treating molecules as such may lead to errors," he says. "Our molecular worms are able to change shape during the traversing of a chemical labyrinth, which allows them to reach areas not accessible to either a single large spherical probe or a rigid real-shape probe. This significantly extends the range of probes and structures that can be efficiently examined."
As a molecule navigates through a chemical system, its access to a particular site or place within that system determines the extent to which catalysis and other chemical reactions may occur. Many of these critical sites are either buried in clefts, pockets or hidden cavities, or else represent channel systems. The accessible volume of a chemical system – the free volume available to a penetrating molecule - is also critical to the system's physical properties, including diffusion, viscosity and electrical conductivity. Predicting whether a molecule will be able to traverse through a given chemical labyrinth is the first question that a simulation must answer, followed by identifying the shortest transverse route, finding the largest probe that can transverse though the system, and calculating accessible volume.
"The required calculations become quite expensive as one needs to include interactions of all the atoms of the penetrating molecule with all the atoms in the labyrinth, and this procedure has to be repeated at every step of the simulation," Haranczyk says. "Additionally, in molecular dynamics only one trajectory per molecule is investigated. Since a penetrating molecule can be bouncing off the walls of a system before it finds a way out, mapping accessible volume in a chemical labyrinth may require the running of a very long simulation to actually see the molecule moving through."
Haranczyk, looking to automate the process by which the void spaces of porous materials are analyzed, had an idea for a probe that would walk through the inside a material and map it. Sethian had been working on mathematic techniques that can be used in robotic navigations and path planning, as well as a host of algorithms for computing geometries in complex settings.
"What's exciting here is to bring together two disparate worlds to build a new technology" says Sethian. The two scientists pooled their expertise to develop the molecular worm algorithm, which they first tested on a zeolite material. Zeolites are microporous minerals that have been widely used since the late 1950s as chemical catalysts, membranes for separations, and water softeners. They are especially useful as alkane-cracking catalysts in oil refinement.
"There are 190 zeolite structures known to exist today, but they constitute only a very small fraction of the 2.5 million structures that are feasible on theoretical grounds," Haranczyk says. "The development of a database of hypothetical zeolite structures has long been regarded as an important step toward designer catalysts as it could, in principle, be screened for zeolites of any property. However, brute-force screening of all possible zeolite structures through molecular dynamics characterization is computationally infeasible, hence the need for rapid triaging based on an initial analysis of various properties."
The successful testing of the molecular worm algorithm on a typical alkane-cracking zeolite opens an immediate door to its use in screening for new zeolites as well as a wide variety of other porous materials. The algorithm should also prove valuable in the search for materials that can capture carbon emissions before they enter the atmosphere. With further refinements, it could also one day be applied to proteins, especially enzymes.
"Being at the frontier of science and solving a very complex problem that has not been addressed before is always very exciting," Haranczyk says.
Cisco has announced new technology that supports its Data Center 3.0 strategy to help customers increase the flexibility of their data centers as they become more virtualized and cloud-based. The new technology advances Cisco's underlying unified fabric capabilities that help customers enhance the efficiency of information delivery in physical and virtualized data center environments, and manage public and private cloud resources more effectively.
"The F-Series modules on the Cisco Nexus 7000 series are currently deployed in LLNL's high performance computing infrastructure, offering us a high density 10GE and low latency networking solution. This technology has enabled LLNL to build large storage network fabrics to support the world class supercomputing systems vital to the laboratory's national security research and development missions," said Matt Leininger, deputy for advanced technology projects at Lawrence Livermore National Laboratory.
"NASA's Nebula cloud computing project based at Ames Research Center, Moffett Field, Calif., is using Cisco's FabricPath Switching System to develop a method to connect local virtualized servers to other cloud networks. NASA also uses the system to interconnect local cloud development and research networks through the Ames Internet eXchange (AIX)," said Ray Obrien, Nebula project manager at NASA Ames.
Announced today is Cisco FabricPath, networking technology that dramatically increases network scalability, resource agility, asset efficiency, and performance in the data center. Cisco also announced new enhancements for Cisco Nexus and Catalyst data center switching platforms, Cisco Wide Area Application Services (WAAS) extensions, and new Cisco services. Comprising the richest set of networked data center solutions in the industry, Cisco's data center and virtualization vision combines unified fabric and unified computing to provide a foundation for reliable, efficient, agile, and highly secure data centers.
Application Performance Optimization
Cisco WAAS accelerates application traffic over the wide area network, enabling enterprises to consolidate applications into data centers and utilize cloud computing, while ensuring performance and productivity for users in remote sites or on the go.
Force10 Networks has announced that it recently provided the networking infrastructure, including an E-Series core switch/router, C-Series resilient switch/router and an S-Series switch as well as engineering expertise, to the recent 26th Chaos Communication Congress (26C3) in Berlin, Germany. The annual four-day conference, organized by the Chaos Computer Club (CCC), offered lectures, workshops, project presentations and other activities to more than 3,000 attendees and some 9,000 remote participants.
“Since 1984, the Chaos Communication Congress has served as the pre-eminent destination for hackers in Europe to exchange ideas and work on projects that help them better understand issues related to computer technologies, security, society and networking,” said Elisa Jasinska, CCC. “We are glad Force10 provided us with high performance network equipment and valuable technical support so we were able to keep the activities operating smoothly throughout the conference.”
In addition to the lectures and workshops, the 26C3 conference featured a hack center, which is considered a huge laboratory for technological research in a wide range of fields, including operating and testing network hardware and software. Drawing from a long-term successful cooperation between the CCC and Force10, the CCC selected the Force10 E600 as a single core switch and the C300 and S50 solutions in the distribution layer and implemented them to anchor the network infrastructure for the conference and the hack center activities.
“As a technology-driven organization, Force10 strongly supports the primary mission of the 26C3 conference, which is to cultivate and share knowledge with concern to computer and network technologies and security,” said Marc Bruyere, Senior Customer Support Engineer, Force10 Networks. “Because we take this effort very seriously, we were excited to provide resources in equipment and expertise to best support the CCC and the work being done during the conference.”