BusinessWire
Nov 15, 2006 09:00 ET

Purdue University Explores SGI RASC Technology to Lower Costs in Scientific Production Environments

FPGA Initiative Uses SGI High Performance Compute and Storage Systems to Speed Mitrionics BLAST Genomics Codes and Operate Science Portals More Efficiently

TAMPA, Fla., Nov. 15 /PRNewswire-FirstCall/ -- SUPERCOMPUTING 2006, Booth #905 -- How do you run more code for more users and lower your electric bill at the same time? How do you economically serve science gateways with tens of thousands or perhaps a hundred thousand users? Purdue University seeks to answer those questions using recently purchased SGI (NASDAQ:SGIC) high performance compute and storage technology from the SGI(R) Scientific Workflow Solutions portfolio.

Purdue University's Rosen Center for Advanced Computing (RCAC) has started a Field Programmable Gate Array (FPGA) initiative specifically attuned to a scientific production environment. They will test genomics and other codes and explore cost-saving computing strategies using the scalable SGI(R) Altix(R) 450 server configured with groundbreaking SGI(R) RASC(TM) (Reconfigurable Application-Specific Computing) RC100 Blades for FPGAs to be installed in the fourth quarter of this year.

At Purdue's West Lafayette, Ind., campus, the SGI Altix 450 system with two SGI RASC RC100 Blades enables four FPGAs, working in concert with four dual-core Intel(R) Itanium(R) 2 processors, to tap into an Altix system's shared memory and address space at full memory bandwidth. The FPGAs in the system can be reconfigured via software to accelerate different applications at different points in time, providing up to 100x performance improvement. While Purdue also purchased a SGI(R) Altix(R) 4700 system and 30TB SGI InfiniteStorage array for immediate production use -- including science portals, nano-hubs, and grid computing -- the FPGAs will initially be used as a separate development environment focused on code porting, production and operations. Exploring the reduction of costs of science portals was a major consideration in the purchase of SGI RASC technology.

David Braun, Research Programmer, Rosen Center for Advanced Computing, Purdue University, has chosen to begin with Mitrionics Mitrion-C and Mitrion Virtual Processor to accelerate applications from multiple disciplines, starting with an accelerated version of the NCBI BLAST bioinformatics application designed to run on the Mitrion Virtual Processor within the SGI RASC RC100 computational blade for SGI Altix servers. BLAST is used by scientists and researchers worldwide for similarity searches for genes and proteins and is the main tool for data mining of large databases in molecular biology. The Mitrion-accelerated BLAST will be available by the end of the year and, under Mitrionics' preliminary testing, is achieving performance increases up to 20X. Braun is also interested in experimenting with other codes in nanotechnology, image processing, and high-energy physics and is currently recruiting researchers who are interested in exploring the possibilities of FPGAs with SGI RASC technology.

"SGI RASC is the operating system support for the FPGA blades and one of the things that I like about RASC is the idea that the operating system can participate in the management of the FPGA: that's what makes it a lot easier for the user," said Braun. "On the management side, the program allows you to reserve an FPGA so that no other program can come in there and reprogram it. From an operational standpoint I'll be looking at power consumption and heat. We already know that, out of the box, an FPGA draws one-quarter to one-third the power of a CPU. That reduces both the light bill and the thermal load. If you have a BLAST implementation, for example and you put up a portal, all of a sudden you have 100,000 people using it and you can keep that machine fully busy just doing that. At that point, your operating costs are going down. You're taking existing science and making it scalable to the masses in an economical way."

According to Braun, just one of the computer systems in the RCAC costs an estimated $200,000 a year in electricity -- and that's with a discount because the University generates some of its own electricity. Operating under a yearly fixed budget from the state, if FPGAs could scale RCAC costs back by just 10 or 20 percent every year, the benefits would mount quickly. SGI RASC can also cut costs by allotting data between CPUs and blades. Some large problems are typically run on MPI or the shared memory processing of SGI Altix systems. If a problem comes in and it's determined that it really needs to have 4 CPUs to do the job correctly, Braun suggests it can be cut by half, to maybe 2CPUs and 2FPGAs doing the equal amount of work. "It may not be much of a speedup per se, but the cost of running that code has gone down," said Braun.

In conclusion, Braun said, "It's an interesting endeavor and I'm glad I'm actually part of it because I think a lot of things are coming down the pike. For example, you can have a FPGA system such as the Altix and the RASC blades actually running the database queries. An Oracle database could sit in the front and submit things down to the FPGA, to keep the sequel statement, and sift through the data in a very efficient manner. I enjoy doing something new like this, something that's not mainstream. I think it's bold to take a step and say 'Lets try something new.'"

"Mitrionics and SGI worked together very closely to develop this turnkey bioinformatics solution for FPGA Supercomputing," stated Anders Dellson, CEO of Mitrionics, Inc. "We're extremely pleased with our relationship with SGI and by the early customer interest in our joint solution. This marks the first milestone in our Mitrion Open Bio Project, where we will continue developing accelerated applications for the bioinformatics industry, and then make them available to the community as open source."

"Purdue University is taking the lead in exploring the exciting new SGI RASC technology from an operation and production standpoint and we are confident they will reduce their cost of operations even as they discover enhanced production speed and efficiencies," said Michael Brown, sciences segment manager at SGI. "Unlike competitors' blade-based systems with a set number of CPUs, I/O and memory, SGI Altix blade servers give customers the choice to mix and match CPU blades, FPGA blades, memory blades and I/O blades to meet their unique computing needs. This increased system flexibility results in more unique research, higher impact results."

SGI and Mitrionics are debuting an accelerated version of the NCBI BLAST bioinformatics application designed to run on the Mitrion Virtual Processor within the SGI(R) RASC(TM) RC100 computation blade for the SGI Altix family of servers at Supercomputing 06, Booth #905.

SGI - Innovation for Results(TM)

SGI 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., and can be found on the Web at www.sgi.com.

NOTE: 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. 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.

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.