Tight Node Connectivity and Compute Power of SGI Altix System Helps Scientists Solve Classic Equations and Develop Powerful Quantum Methods
MOUNTAIN VIEW, Calif. (November 27, 2006) — What if, before designing a car, an engineer could use physical knowledge of atomic particles to design materials and know exactly how they would react in a collision? What if fundamental physical equations that predict how those particles will behave in various energy states (cold, heat, stress) could be used to design spacecraft? To solve the physics equations that will lead to the engineering designs of tomorrow, physicists at the University of Rhode Island are investigating the fundamental quantum mechanics of particles using technology from SGI (NASDAQ: SGIC). The SGIÂ® AltixÂ® system, installed in August, is used by the Physics Department to solve one of the basic equations of quantum mechanics, the SchrÃ¶dinger equation, which describes how matter behaves at the atomic scale. While the equation has been solved in several simple cases, SGI Altix technology was chosen because systems involving more than two or three particles cannot be directly solved, and require instead the use of computationally intensive numerical methods.
"We selected the Altix because of the speed of the connectivity of the various nodes, the speed of the exchange of information between the nodes, the speed of the individual processors, and the flexibility that that offers," said Dr. Peter Nightingale, professor of physics, University of Rhode Island. "That is important because a lot of what we do involves linear algebra with matrices, and the matrices tend to be spread out over different nodes. There is communication necessary to do a coordinated calculation for a matrix that is spread out over these systems. It’s the speed of that communication that really is a bottleneck at times. Our cluster system was simply becoming unreliable and the SGI Altix offers much stronger connectivity and therefore we are able to do bigger calculations faster, which require a lot of compute power."
Purchased in July through James River Technical, Inc. (JRTI), SGI’s exclusive higher education reseller, the SGIÂ® AltixÂ® 350, with 10 IntelÂ® ItaniumÂ® 2 processors running NovellÂ® SUSEÂ® LinuxÂ® Enterprise Server 9, is connected to the older cluster as well as to numerous desktops in the Physics Department. The SGI Altix system is also connected to the Internet, allowing anyone with access to use it, typically students who log in from home.
The SGI Altix 350 system will be used by Dr. Nightingale and his students as part of an ongoing program, supported by the National Science Foundation (NSF), to study the behavior of small van der Waals complexes and to develop their own applications. These van der Waals systems consist of a small number of weakly interacting atoms, and the research addresses the fundamental problem of solving the SchrÃ¶dinger equation for these systems. The University of Rhode Island researchers will develop new Monte Carlo methods to solve this equation, and the SGI Altix system will supply the power to speed up study of these particle systems and compare them with experiments. As Dr. Nightingale explains, the development of these methods also has implications for future engineering design.
"The energies of ground and excited states tell us about how the particles interact with each other, but what is not really known is how complicated systems interact, what the strength of the interaction is, as the distance changes," added Dr. Nightingale. "For instance, if you can accurately predict where the energy levels are, you can figure out what the interactions between the particles are and that is important for all sorts of applications. Ultimately, and we may not be there for quite a while, the idea is to write down these fundamental physical equations and then design materials on the basis of the fundamental properties of this material. Right now, engineers design products based on what are called phenomenological models, which are a mixture of things that are known, things that are guessed, and things that are measured. But it would be much more elegant if you could start from fundamental physics, from the SchrÃ¶dinger equation, and on the basis of that predict how your car will behave in a collision, for example. That’s quite a stretch, but that’s the ultimate goal."
"Quantum mechanics applications are among the most computationally intensive, and are critical for the advancement of many areas of physics, nanotechnology, material science and drug discovery," said Michael Brown, sciences market segment manager, SGI. "The University of Rhode Island’s Physics Department chose the SGI Altix system for its combination of speed and scalability, allowing them to develop new, breakthrough approaches to highly accurate simulations. We have also formed a partnership, as we do with many scientists and researchers who are pushing the envelope of computation, in which URI and SGI agree to explore ways in which SGI architecture could help speed up calculations even further."
SGI - Innovation for Results™
SGI (NASDAQ: SGIC) delivers a complete range of high-performance server and storage solutions along with industry-leading professional services and support that enable its customers to overcome the challenges of complex data-intensive workflows and accelerate breakthrough discoveries, innovation and information transformation. SGI helps customers solve their computing challenges whether it’s enhancing the quality of life through drug research, designing and manufacturing safer and more efficient cars and airplanes, studying global climate, providing technologies for homeland security and defense, or helping enterprises manage large data. With offices worldwide, the company is headquartered in Mountain View, Calif.
© 2006 SGI. All rights reserved. SGI, Altix, the SGI cube and the SGI logo are registered trademarks of SGI in the United States and/or other countries worldwide. Linux is a registered trademark of Linus Torvalds in several countries. Novell is a registered trademark, and SUSE is a trademark, of Novell Inc. Intel, Itanium, and Xeon are trademarks or registered trademarks of Intel Corporation or its subsidiaries in the United States and other countries. All other trademarks mentioned herein are the property of their respective owners.