INDUSTRY

SGI Technology Accelerates Performance by Up to 1,000 Times, Fueling Breakthroughs in Engineering, Bioinformatics and Computer Sciences: At the University of California, Riverside, faculty and administrators understand the enormous value of collaborative research, in which investigators in one discipline work closely with researchers in other fields. Such an approach can speed innovations that sharpen the technical edge of the university, of industry, and of America itself. Collaborative research is nothing new to UCR, which has a longstanding program of interdisciplinary studies and close partnerships with industry. But with a powerful new supercomputer from SGI under its roof, the university's efforts are moving into overdrive. UCR's new SGI Altix 4700 supercomputer -- the largest single Altix 4700 computer in the University of California system -- is powered by 64 Intel Itanium 2 processor cores and features 128GB of system memory. The acquisition is the result of an effort led by UCR's Department of Computer Science and Engineering, with the purchase funded entirely with a grant from the National Science Foundation. Unveiled today in a press conference at the UCR Bourns College of Engineering, the SGI Altix system, which was installed in December 2006, will serve as an essential resource for a new generation of studies in computer-based modeling and simulation, Internet architectures, bioinformatics, machine learning and data mining. "This system is a good example of how SGI delivers solutions for the most demanding compute and data-intensive workflows," said Dr. Eng Lim Goh, Senior Vice President and Chief Technology Officer, SGI. "The flexible, open and scalable Altix architecture allows the right combination of compute, special purpose processing, memory and I/O elements, to meet the varying needs from fundamental to multi-disciplinary research." "This is a very important and strategic acquisition for the University and for the UC research community," said UCR Chancellor France A. Cordova. "In its 52 years, UCR has evolved into a leading research institution that is unique among Southern California universities. Today we turn to the next chapter -- one written by technology pioneers inspired by a spirit of discovery and fueled by a passion for innovation." An expandable system to meet escalating needs UC Riverside's rapidly growing array of research projects prompted the campus to acquire a new Linux system capable of handling larger, more complex problems than traditional cluster computers could manage. Rather than acquire a series of less powerful systems devoted to specific projects, UCR's researchers and administrators were led by the Department of Computer Science and Engineering to adopt a more strategic approach: securing NSF funds to purchase a large system that can be shared with other departments, and even with other universities. The flexible and open Altix system architecture can accommodate multiple users and applications at once. For instance, one researcher might access a large share of the system, while splitting the remaining system memory and processors between several other users. Or the entire system could be applied to solve a single, particularly complex or data-heavy problem. To software, the Altix presents a single system image, which makes it significantly easier to program and administer than a typical fragmented Linux cluster. And by running the entire system under a single copy of Linux -- in this case, Novell SUSE Linux Enterprise Server 10 -- UC Riverside researchers can more easily manage their projects on the Altix. In contrast, typical clusters run a copy of Linux on each node. This can complicate project management for engineers and scientists who otherwise would prefer to concentrate on their research. "We selected SGI Altix because it meets the current needs of the particular NSF project and those of our full spectrum of engineering and science researchers, while offering plenty of 'headroom' for more extensive workloads in the future," said Laxmi Bhuyan, Professor of Computer Science and Engineering at the Bourns College of Engineering, which houses the new Altix system. Professor Buhyan, Project Director and Principal Investigator for the NSF grant, added that "with a single resource that is far easier to administer than a cluster running multiple instances of Linux, UC scientists and engineers can focus on their work -- not on how to get their applications to run on the system. With this new platform for discovery, we anticipate realizing tremendous gains in productivity, as well as important new breakthroughs in a broad range of studies." The "headroom" that UCR enjoys with this new system is substantial, since Altix 4700 systems can support up to 1,024 Intel Itanium 2 processor cores and up to 6TB of memory running under a single copy of Novell SUSE Linux Enterprise Server 10 -- making it by far the most scalable platform in the industry. SGI RASC blade accelerates work a thousand-fold Another prime reason for acquiring the SGI Altix system for UCR's Department of Computer Science and Engineering was the availability of SGI RASC (Reconfigurable Application Specific Computing) technology. SGI's RASC technology uses powerful Field Programmable Gate Arrays (FPGAs) to allow the computer to execute multiple functions at once. The SGI RASC RC100 blade accelerates the performance of applications that spend most of their time running a small number of routines, such as pattern-matching algorithms that search massive genomic databases to find similar genetic sequences. Professor Najjar's team has developed software tools that make it easy to program FPGA technology for specific uses. The team found that the FPGA technology used in SGI RASC solutions executed some important scientific code, such as the Smith-Waterman algorithm, as much as 1,400 times faster than a typical high-end desktop system. To run its specialized software tools on the RASC blade, UC Riverside also purchased an SGI Altix XE240 system powered by four Intel(R) Xeon(R) processor cores and 16GB of memory. Because the tools are not designed to run on Intel Itanium 2 processors, the SGI Altix240 server is an ideal base node for applications written for x86-64 processor platforms. "The combination of SGI Altix 4700 servers and Dual Core Intel Itanium 2 processors results in a platform that is fast, stable, efficient and enormously scalable," said Richard Dracott, General Manager, High Performance Computing, Intel. "The UC system's collaborative research community can confidently view this as a foundation for exciting new discoveries." Facilitating the acquisition of SGI solutions was James River Technical, Inc. (JRTI), SGI's partner in delivering high-performance computing solutions to the higher education and research sector. "UC Riverside understands not only the benefits of collaborative research, but also the tremendous economies of scale that come from investing in a larger, more powerful shared resource instead of a series of free-standing servers used for specific projects," said Tom Mountcastle, President, JRTI. "The university's vision will no doubt bring significant advances not just to its research user base, but to the UC system and the Southern California region as a whole. This is a strategic procurement for UCR and a tremendous win for our team in the California university system."