North Texas Daily

Matthew Brown / Intern Writer

Tucked away on the fifth floor of the General Academic Building, there is a room housing rows of servers that comprise the most powerful computing system at UNT, the school’s very own supercomputer.

The supercomputer, the Talon 2.0, is used mostly for research and carries out complex funtions in a fraction of the time a normal computer would require.

“Not every researcher has funding to buy equipment, so this provides a stable amount of computing for a large part of the university,” High Performance Computer Services Manager and chemistry adjunct Scott Yockel said.

Fifty five research teams and counting from fields such as biology, chemistry and mathematics are using Talon to simulate complex models. The supercomputer, and the research it facilitates, is part of the push to make UNT a nationally recognized research university.

“I just had a paper accepted yesterday in one of the top journals in coding theory,” associate mathematics professor Anne Shepler said. “I probably wouldn’t have started the coding theory project if I weren’t aware that Talon was on campus.”

Most home and school computers use between 4 and 8 GB of RAM, while the most powerful consumer computers max out at 32 GB of RAM. The Talon is divided into clusters and then into nodes that have varying amounts of RAM. Combined, the Talon has more than 14,000 GB to accomplish any challenge presented to it and up to 512 GB available to any single task.

“I was doing some calculations and I was able to run the code through Talon and get it done in a couple minutes,” doctoral student Christine Uhl said. “On my desktop it would have taken a couple hours.”

The Talon 2.0 was brought online last fall when it replaced the 4-year-old Talon 1.0.  The new system has approximately five times more processing power than the first and is better suited for intense computations, such as modeling complex molecular systems.

The system’s yearly budget is just short of $970,000, which has turned out to be a savvy investment. Yockel said that grant money has tripled since the first Talon was brought online, from a $1 million a year to more than $3 million a year.

“If we didn’t have a system like [Talon], researchers would build smaller systems in their labs and they would have much more limited use,” Yockel said.

UNT’s Board of Regents has agreed to upgrade the system every three years to keep the computer competitive. Until its next upgrade, the Talon 2.0 will be the most powerful computing resource available to the growing number of researchers using the system, which is expected to reach 500 by the end of 2016.