Penn State University Selects Unisys ES7000

BLUE BELL, Pa.--Unisys Corporation today announced that The Pennsylvania State University's Information Technology Services (ITS) organization has selected Unisys ES7000 enterprise servers for use in its high-performance computing (HPC) environment. The technology will help quantify the benefits of using Intel's 64-bit Itanium 2 processor architecture in computing tasks covering a broad set of academic disciplines. These include computational biology, chemistry, fluid dynamics, gravitational physics, mathematics, materials science and structural mechanics. In addition, the technology will be used to support ITS' visualization initiatives for representing business or scientific data as images that can aid in gaining better insight from the data. The ES7000/430 system will be made available to Penn State's faculty and students by the Graduate Education and Research Service (GEaRS) group, a unit of ITS at Penn State. "We believe there are important price-performance benefits that will make scale-up computing on the Intel Itanium 2 processors, such as those offered by the Unisys ES7000/430, the dominant 64-bit architecture over the next few years," said Vijay K. Agarwala, director, GEaRS. "Our ES7000/430 will be particularly useful for researchers who need more than four gigabytes of memory per process and those working with large database applications." The centerpiece of the installation consists of a Unisys ES7000/430 with two domains of 16 Itanium 2 processors at 1.5 GHz each. This clustered technology will provide an alternative approach to how GEaRS currently meets the high-performance computing needs on the campus with over 600 servers containing dual 32-bit Intel Xeon processors MP. Key to the university's choice of server technology was the access to large amounts of shared and high-bandwidth memory that Itanium 2 and the ES7000 will provide. The university will help demonstrate the merits of Intel-based "vertical scaling" -- running applications in a single, expandable environment -- for high-performance computing solutions. Those advantages include the ability to support large databases for applications with data structures that cannot be partitioned easily; an alternative to more labor-intensive methods of load balancing; the ability to use a single operating system image; and simpler systems administration. "By combining the superior economics of commodity component technologies with mainframe-class server design, Unisys, Intel and Penn State are showing how to drive down the cost of high-performance computing to the point where it can be an effective research tool for organizations who once could not afford it," said David Houseman, vice president, Unisys Advanced Technology.