Virginia Tech Adds SGI Altix 4700 for Large-scale Chemistry Applications

SGI Shared-Memory Architecture Cuts Run Time of Calculations in Half

MOUNTAIN VIEW, Calif. (September 12, 2006) — To primarily run Gaussian for chemistry research and development, among other code, Virginia Tech again selected technology from SGI (OTC: SGID). After using the shared memory architecture of an SGI® Altix® system for a year, the university purchased a new SGI® Altix® 4700 and upgraded their existing system by doubling the amount of processors. Approximately 80 percent of the CPU cycles on the new SGI Altix system will run Gaussian calculations. These include simulations to understand how atomic oxygen can damage materials in low earth orbit and for research in the development of molecules for new drug discoveries. The Chemistry Department also writes a good deal of their own in-house code, and the ease of code adaptation to the shared memory, Linux® environment of the SGI Altix system was an important feature in their purchase decision. In addition to chemistry, both SGI Altix systems are being used for research in geosciences, aerospace, chemical engineering, and a variety of other disciplines requiring the utmost in high performance and speed.

"The thing I have been most pleased with on the Altix is the shared memory with a very fast interconnect," said Dr. T. Daniel Crawford, associate professor, chemistry, Virginia Tech. "Doing fairly large scale, threaded calculations, the speed of the interconnect and the proximity of local memory means that you can run significantly larger calculations, and do it more easily than on distributed memory computers. Software development is easier on a shared memory system like the Altix and therefore you can get more work done in a shorter period of time."

While Dr. Crawford feels that distributed memory systems are very important, software has not kept pace with hardware, "not even by a factor of 10, I would say." He explains that distributed memory systems require a fundamental rewrite of home-grown programs to take advantage of the hardware, which takes a lot of time, and is a big activation barrier to surmount. "The difference between a supercomputer like that and the Altix is that you can get over that activation barrier much faster," he said. "You can start doing large-scale calculations really quickly with the Altix because it doesn't take as long to modify your software."

Dr. Crawford and his group are studying chiral molecules for natural products chemistry (where compounds are taken from natural sources like tree bark, marine animals, etc.). One of the most important potential applications of this work is in trying to assist in the development of new chiral drugs. Dr. Crawford notes that the SGI Altix system is particularly fast for Gaussian apps in this area.

"The threading capabilities of a program like Gaussian are impressive," said Crawford. "I think Gaussian has done a very good job with their threading code so that it can really take advantage of the Altix's shared memory environment. You can run a calculation, requesting, say, 12 processors, and while you may not get linear speed up, you will see a dramatic speedup as opposed to just using a single processor."

Dr. Crawford and his group's home-grown codes are on the very high-accuracy end of quantum chemical models. (The programs are open-source and freely distributed under the GNU GPL at http://www.psicode.org) Depending on the size of the molecule, there are relatively huge data files. It's not unusual for the group to use 300GB of disk space for one calculation. One of many advantages the researchers have found with the SGI Altix system is that it has such a large shared memory pool — on the order of well over 300GB of aggregate memory available — that, coupled with the Linux operating system, which uses buffered I/O, a large calculation can run at nearly 100 percent CPU efficiency.

Crawford added, "The idle time on the Altix is very low and I can run a large-scale calculation that takes, say, three weeks on one of my Opterons, in about a week and a half on the Altix. I don't have to wait on the disk all the time."

Ordered in April and delivered in June, Virginia Tech purchased from James River Technical, Inc., SGI's exclusive higher education value added reseller, an SGI Altix 4700 system with 64 Intel® Itanium® 2 processors that is currently in the process of being upgraded to SUSE Linux Enterprise Server 10 and ProPack5. At the same time, the university upgraded their existing 64-processor SGI® Altix® 3700 Bx2 to 128 Intel Itanium 2 processors. Virginia Tech is also in the process of upgrading their 2TB of scratch disk, adding an additional 3TB.

Tom Mountcastle, President of JRT says of these purchases, "Virginia Tech has continued to be a strong partner with SGI and JRT, and clearly understands, as illustrated by Dr. Crawford's results, the benefits offered by the Altix Itanium systems. We look forward to continued growth in their systems in the coming year. They continue to be true innovators."

"With their first SGI Altix purchase a year ago, Virginia Tech discovered the powerful, performance-enhancing capabilities of SGI global shared-memory architecture," said Michael Brown, sciences market segment manager, SGI. "They wanted ease of use, they wanted to run both serial and parallel codes and they wanted reliability — all of which SGI technology delivers throughout the academic and commercial scientific research community."

SGI | The Source of Innovation and Discovery.™
SGI, also known as Silicon Graphics, Inc. (OTC: SGID), is a leader in high-performance computing. SGI helps customers solve their computing challenges, whether it's sharing images to aid in brain surgery, designing and manufacturing safer and more efficient cars and airplanes, studying global climate, providing technologies for homeland security and defense, enabling the transition from analog to digital broadcasting, or helping enterprises manage large data. With offices worldwide, the company is headquartered in Mountain View, Calif., and can be found on the Web at www.sgi.com.

This news release contains forward-looking statements regarding SGI technologies and third-party technologies that are subject to risks and uncertainties. These risks and uncertainties could cause actual results to differ materially from those described in such statements. The reader is cautioned not to rely unduly on these forward-looking statements, which are not a guarantee of future or current performance. Such risks and uncertainties include long-term program commitments, the performance of third parties, the sustained performance of current and future products, financing risks, the ability to integrate and support a complex technology solution involving multiple providers and users, and other risks detailed from time to time in the company's most recent SEC reports, including its reports on Form 10-K and Form 10-Q.

Silicon Graphics, SGI, Altix, the SGI cube and the SGI logo are registered trademarks, and The Source of Innovation and Discovery is a trademark of Silicon Graphics, Inc., 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. in the United States and other countries. Intel and Itanium 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.