Washington University expanding supercomputer with NIH grant

Washington University in St. Louis has received a nearly $600,000 grant from the National Institutes of Health to upgrade and expand the university’s existing high-performance computer that will allow researchers to process large amounts of data from their research significantly faster.

Along with matching funds from the School of Engineering & Applied Science and the Mallinckrodt Institute of Radiology at the School of Medicine, the total funding is nearly $1 million.

Fred W. Prior, professor of radiology, and Rohit V. Pappu, professor of biomedical engineering, received the one-year, $597,700 grant to add new hardware to analyze the massive amounts of data being created by university researchers. The existing computer was purchased with a $2 million grant awarded in 2009.

“This has the potential to really highlight how, in highly relevant, transformative biomedical problems, computations are really driving discovery,” says Pappu, also director of the Center for Biological Systems Engineering.

Researchers and engineers throughout the university will use the supercomputer, housed at the School of Medicine, in four main areas:
  • Analyzing image data to map the Human Connectome;
  • Analyzing genomics data to uncover genomic bases of disease;
  • Extracting biomedically relevant insights by analyzing large amounts of data;
  • Understanding how molecules and their dynamics control decisions of cells.
“Research computing requires high-performance computing infrastructure,” Pappu says. “The way we can get ahead of the pace and the demand is to fundamentally change the architecture that was being called upon. We always need to push the envelope.”

Prior, co-director of the Center for High-Performance Computing (CHPC) with Pappu, says the current high-performance computer is on the highest bandwidth network at Washington University. To date, the CHPC has processed more than 3.4 million jobs that have consumed a total of 19.8 million processor hours.

“And now there’s a high-speed link dedicated for research between the campuses, and we’re hoping to maintain and grow that,” Prior says.

The new supercomputer is built with GPUs, or graphics processing units, similar to what are used in computer games. The GPUs are made up of thousands of smaller, more efficient cores designed to handle multiple tasks simultaneously, which helps to speed up processing, Pappu says.

“What today takes about three weeks to process and soaks up quite a few of the nodes on the current high-performance computer should take three days or maybe a day,” Pappu says. “But now that we have this GPU-based system, what we are going to ask of it will exponentially increase.”

Matching funds also allowed Prior and Pappu to hire Jiaofeng Xu as a dedicated programmer to help the research community with the supercomputer. Xu, who earned a doctorate this month in biomedical engineering, worked in the lab of Mark Anastasio, professor and interim department chair of biomedical engineering.

Both Prior and Pappu plan to keep looking ahead at the needs of the research community to keep the supercomputer updated. Pappu says the speed of processors basically doubles every 18 months, which requires ongoing investment.

“We understand at any given time how much you need to invest and that you have to be clever about what you’re investing in, then recognize that you need to come back 18 months later and reinvest,” Pappu says. “That’s a good place to be.”