INDUSTRY

Indiana State University is joining the ranks of high-performance computing. University officials announced today the acquisition of a powerful computer cluster capable of performing nearly a trillion operations per second, a capability that will greatly improve the university’s standing as a research institution. “It really changes the landscape for higher education in this state,” Indiana State President Lloyd W. Benjamin III said of the acquisition. “We are the fourth [institution] in the state to acquire computing capacity of this dimension and this size, and that has implications for Indiana State.” “It’s kind of like going into Divison I athletics; it puts you in that category,” Benjamin said. “And when you are in that category of institution and attempting to recruit new players -- our players being faculty and graduate students * when they understand the capability that is resident here at Indiana State University, it’s going to help us not only retain faculty but also recruit a caliber of faculty who in their day-to-day work are going to rely upon this being in place.” The system consists of 32 interconnected computer nodes with a combined total of 64 AMD Opteron processors. Each 64-bit processor clocks operation speeds of up to 2.8 gigahertz per second. “It’s a personal computer on steroids,” says Kenneth Janz, director of instructional and research technology services at Indiana State. “Instead of having a single processor or even a dual processor that’s in a basic workstation, now we have 64 processors and we have a lot more RAM, we have a lot more disk space * Everything about it is high performance. The idea is to get as many computational cycles out of that machine as possible.” Installed at a cost of $172,000 through the use of quality improvement funds, the system’s 10.96 terabytes of hard drive storage can be expanded even further through access to a fiber channel array. With extreme high-speed processing and robust memory, high performance computers are used for a variety of specialized applications that require huge amounts of sophisticated mathematical calculations. As the use of technology has increased in academia, computational simulation has joined theoretical analysis and physical experimentation as a tool for discovering and investigating new information. Indiana University, Purdue University and the University of Notre Dame are the other institutions in Indiana to have similar computing power. The massive computing power of Indiana State’s new system will open new doors of computational investigation and analysis for university faculty, graduate students and undergrads, as well as external researchers, and greatly reduce the amount of time required to conduct complex, high-end computations. Projects that previously took weeks can now be reduced to a single day’s work. Eric Glendening, a computational chemist at Indiana State, said it’s not unusual for the Indiana State chemistry department’s current 14-processor system to chew on a complex calculation nonstop for three weeks or more. The majority of Glendening’s research entails simulating chemical reactions or properties of chemical systems. “I have two projects that essentially had to stop in my own lab here because I didn’t have the computing power to treat those systems. This [new computer] will allow us to proceed,” Glendening said. “This is essentially a collection of 64 high-end PCs in a single cluster. We’re going to be able to look at more high-end computations than we’ve been able to handle in the past,” he said. “In one day we can do four weeks worth of calculations. Calculations that simply were not possible before will now become possible to us.” A two-processor computer assembled several years ago from then state-of-the art computer components is still in use in Indiana State’s life sciences department. It often runs for weeks on a single computation, as does a similar computer in the physics department. “It’s not just one or two days * it’s months sometimes,” notes physicist Guo-Ping Zhang. And sometimes, Zhang points out, there are power outages. When that happens, the researcher must start over from scratch. Zhang specializes in the fields of nanoscience and computational material science, areas he says are “very, very challenging without this type of computing.” In terms of comparing the new high-performance computer to Indiana State’s current systems, “We’re not talking a comparison of 10-1 or 100-1,” said Charles Amlaner, chair of the life sciences department. “My guess is it’s probably well over 1,000-1. We’re looking at a really, really huge jump in performance, which means you can run more experiments in a shorter period of time.” The addition of a high-performance computer to the scientific toolbox allows research and analysis to a wide array of disciplines that was previously impossible. Without it, Zhang says, “we limit ourselves.” With it, “you can explore something originally you didn’t think of.”