Contact Research

Office of Research

Office of Research & Economic Development
vpresearch@uidaho.edu
phone: 208-885-4989

Physical Location
Morrill Hall 105

Mailing Address
875 Perimeter Drive MS 3010
University of Idaho
Moscow, ID 83844-3010
STEM - proteins

Terabytes of Power

New UI Supercomputer Spits Out Solutions To Problems Previously Too Big to Tackle

By Tara Roberts


A new University of Idaho computer nicknamed “Big-STEM” isn’t much larger on the outside than your average computer tower. But what’s inside puts everyday PCs — and most other computers — to shame.

When Big-STEM goes fully online this summer, it will be one of the most powerful computers of its kind in the United States.

“Basically, Big-STEM means we can solve problems that pretty much no one else is working on,” says computer science professor Jim Alves-Foss, who manages the Big-STEM project for the group of research faculty who will tap into its power to more quickly churn through data sets so complex they would fry lesser systems.

Big-STEM’s power lies in its incredible amount of memory. When it’s complete, Big-STEM will have 8 terabytes of memory – 4,000 times the memory of the average home computer.

A grant from the National Science Foundation funded an initial 4-terabyte phase that enabled UI professors last fall to begin testing its potential. UI received additional funding this month from the Murdock Charitable Trust to add a second 4-terabyte phase.

While other types of supercomputers can handle huge amounts of data, Big-STEM’s big memory allows it to address problems that involve massive interactions among data, such as detailed simulations and models of complicated systems.

“There’s a whole range of complex problems that people haven’t been able to find solutions to because the computing power wasn’t there,” Alves-Foss says. “Some of our faculty members are working on these problems. They’re on the cutting edge of research.”

“This machine gives us the capacity to work on problems that are important to industry, too, supporting Idaho Global Entrepreneurial Mission (IGEM) initiatives sponsored by the Idaho Legislature.”

Big-STEM’s concentrated layout — which when complete will include two machines with a total 320 microprocessor cores — will allow it to quickly and easily process data. Any single program can access all of the memory at once, providing extremely detailed results.

Marty Ytreberg, UI associate professor of physics, is among the faculty members who have already put Big-STEM to work. He studies proteins found in the human body that rapidly change shape.

Many of these types of proteins are implicated in human disease, and understanding them could help researchers design specially targeted drugs. However, one single protein can have as many as 100,000 different shapes.

“It looks like a hairball,” Ytreberg says. “You can’t pick anything out of the structures. They all look different.”

If Ytreberg tried to give even a very powerful common computer the task of finding similarities within the protein structures, the machine would run out of memory and crash. But Big-STEM is up to the task.

“I can do analysis on that computer that I can’t do on any other computer,” Ytreberg says.

In addition to its memory, Big-STEM’s speed has transformed projects that used to take weeks or months to compute.

Fred Barlow, chair of the UI electrical and computer engineering department, says Big-STEM has dramatically sped up the amount of time it takes to create simulations of computer chips and other small electronic devices, and also has allowed researchers to run more detailed simulations.

Barlow and Gabriel Potirniche, an associate professor of mechanical engineering, have used Big-STEM to model thermoelectric devices, which could help harness the energy lost in vehicle exhaust to help improve fuel economy. Such simulations help accurately predict how well different designs will work.

“If every time you make a small change to a design it takes weeks to get an answer, the design process can take a very long time,” he says. “Instead of weeks, Big-STEM can do it in hours.”

Other faculty members have tested Big-STEM in mathematics and other engineering fields — and projects like these are only the beginning.

Alves-Foss says Big-STEM’s abilities put it on par with some of the nation’s highest-performing computing clusters — such as the 8-terabyte Mason system at Indiana University — meaning it offers rare opportunities for research. Alves-Foss is still reaching out to UI professors to find new projects that need Big-STEM’s supremacy.

“It is amazing what we can do with modern computing,” Alves-Foss says. “So often I am reminded that we have brilliance and tremendous capability all across our country, and not just in the Ivy League schools. I am proud to be part of this group and to help our talented young faculty realize the full potential of their research.”