University purchases super-fast computer

The University installed one of the 100 fastest computers in the world last month, becoming one of fewer than 10 universities in the world to own a supercomputer.

"Many fields of science and technology are increasingly dependent upon high-speed and high-volume computing power," President Tilghman said in an email. "This will be a tremendous tool for many of our colleagues."

Now running at 87 Prospect, the system requires five to 10 times less electrical and cooling work than the Beowulf computer clusters the University has used to conduct similar research in the past, said Curt Hillegas, manager of Computational Science and Engineering Support.

The supercomputer also offers a more efficient way for researchers to conduct research. Since the supercomputer's 2,048 processors interact and communicate more efficiently with each other than the processors in the clusters, it conducts calculations three times faster, allowing researchers to create models on a larger scale.

"As problems that are being addressed by research grow, [researchers] have found that one single computer isn't the most efficient way to do things," said Hillegas, who helped select and purchase the system. The supercomputer, one of IBM's family of Blue Gene supercomputers, is the "next step in the evolution of this idea."

Astrophysics professor William Tang noted that the system offers much more than speed.

"It is fast, but that is not such a big, dramatic part of it," Tang said. "What's interesting is that many of the applications, particularly the ones I'm involved in, scale very well."

In other words, the supercomputer allows researchers to work with technology that can connect to even larger processing units at IBM or other national laboratories. "It's like a bridge," Tang said.

Tang, who is also the chief scientist at the Princeton Plasma Physics Laboratory (PPPL), added that the system is especially helpful for fusion research.

"In the research here at the PPPL, we're involved in a very aggressive quest for harnessing fusion energy to address the energy needs of the world, which is very timely right now," he said. "On the simulation side, tools like the Blue Gene/L are very attractive as a platform for advancing science in this area."

Alan Gara, the Blue Gene's chief architect at IBM, also noted that the supercomputer is highly efficient for plasma computations.

"The importance of plasma simulations cannot be overstated now," Gara said. "Interactions with world-class researchers such as those at the PPPL are critical."

Get the best of the ‘Prince’ delivered straight to your inbox. Subscribe now »

Though the PPPL is allowed to use the supercomputer, it is owned by the University, unlike most supercomputers in the world that are owned by or at least affiliated with national laboratories.

The purchase was funded by the Office of Information Technology, Princeton Institution for Computer Science and Engineering and the School of Engineering and Applied Science, as well as five faculty members from the astrophysics, chemistry, chemical engineering and mechanical and aerospace engineering departments, who contributed from personal research grants.

Initially, only those contributing faculty will have access to the supercomputer, but in a few months, it will be available to other researchers by application.

Tang, who is one of the contributing faculty members, noted that the list price for the supercomputer — which is in the millions — is higher than what the University paid for it.

"IBM gave us a very nice academic discount," he said.

Betty Leydon, vice president for information technology and chief information officer at the University, said in an email that the price is confidential, and that the purchase drew significant contributions from multiple sources.

"This shows that we recognize the value of working together to build the best possible IT infrastructure to support research at Princeton," Leydon said.

She added that this "extraordinarily rich research computing environment ... will facilitate collaborative research among faculty and open up new possibilities to researchers who, at this point, may not have thought about how they might use such a resource."

The supercomputer should hold a spot on the list of "Top 500 Supercomputer Sites" when the list is updated next Tuesday.