On December 19th, NACS met with a cross-section of campus faculty “consumers” of High Performance Computing (HPC) to discuss the current state of HPC at UCI. The group included approximately 25 researchers from fields that have historically dominated use of HPC – engineering, chemistry, physics and biological sciences.
HPC refers to significantly faster “number crunching” power than desktop computers are capable of. Contemporary HPC machines can perform hundreds of millions of “flops” (Floating Point Operations Per Second); a typical 200 MHz Pentium PC peaks at about 70 Mflops.
Currently, UCI supports HPC in various ways – by purchasing time at the San Diego Supercomputer Center (SDSC) on the Cray C98 vector machine (7,600 Mflops) and on the Cray T3E massively parallel processor (MPP). NACS also provides access to UCI’s own aging HPC system – a 4 vector CPU Convex C3840 (480 Mflops).
The participants at the December meeting engaged in a lively give-and-take discussion. All agreed upon the importance of the UCI network and of the support given to the faculty and their research group by NACS’ Departmental and Distributed Computing Support group (DDCS). There were differences of opinion about the role of NACS in providing other HPC resources to the UCI community, with some speakers advocating the need for a centrally-managed facility, such as a workstation cluster, or central multi-processor shared memory machine. Others felt that the days of “big-iron” had ended. UCI is not unique in this respect – discussions of this type are going on at many other universities.
If you are interested in participating in discussions about HPC and computational science at UCI, please let us know by sending e-mail to “AAG@UCI.EDU” . For more information on HPC at UCI, please take a look at the following Web page: