News

Supercomputer offers 40-gigabit speeds for university researchers

Owl’s Nest can process massive amounts of data.

Dr. Axel Kohlmeyer types a few lines of code into his Linux computer and an array of red, yellow and white graphs appear on the screen. The program, which he calls his “big brother tool,” helps him remotely monitor the Owl’s Nest – Temple’s first supercomputer.

Kohlmeyer said that the Owl’s Nest is modest as far as supercomputers go.

“It depends on perspective,” he said. “Different people that talk about this computer have seen different size computers. I’ve been fortunate to have experience to have run on extremely large [computers].”

In essence, the Owl’s Nest is a collection of computers hooked to one another.
With a cluster of 180 computers, the Owl’s Nest is approximately 650 times smaller than Oak Ridge National Laboratory’s Titan, the second-largest supercomputer in the world. Still, the Owl’s Nest cluster is a powerful research tool at Temple, enabling scientists to model molecules and weather patterns, study quantum mechanics and analyze genomes.

Kohlmeyer, who is the associate dean of the College of Science and Technology, built the Owl’s Nest in 2011. Today, he and his colleagues maintain the system and train and assist around 200 scientists who use it in their research.

“If you have a fast Ethernet switch at home, you would have a connection that would allow you to send one gigabit per second,” Kohlmeyer said. “The technology that the cluster has allows you to send 40 gigabits per second.”

The cluster allows scientists to process large amounts of data more efficiently. While a single computer might take weeks to process a large dataset on its own, the Owl’s Nest circumvents the problem by dividing large datasets into smaller ones and assigning them to different computers. The computers continue to communicate with each other to prevent inconsistencies.

“If you would do it on a single computer, it could take years or decades, but if you can break it down so that you can run it on hundreds of CPU cores, things become feasible,” Kohlmeyer said.

But computers produce a lot of heat, so the cluster’s computers are arranged on racks that are connected to fans to keep the cluster cool.

“It’s pretty noisy because all the machines have to have lots of fans,” Kohlmeyer said.

“We live in an age of ‘big data,’ all fields have access to big data these days, the chemists, the physicists, and now the biologists,” said Dr. Rob Kulathinal of the College of Science and Technology. “A lot of people do simulations [using Owl’s Nest].”

Kulathinal and his team use simulations to study evolutionary biology.

Craig Stanley, a senior graduate student in Kulathinal’s lab, said the supercomputer helps increase the capability of what Temple’s scientists can now achieve.

“This allows us to really push the edge on how we analyze data,” Stanley said. “Instead of analyzing two genes like I would have done in my master’s [work], we analyze three hundred genomes at the same time.”

But, like all computers, Kohlmeyer said the Owl’s Nest is not immune from becoming obsolete.

“The pressure to buy new hardware is always there because people want to run more [data],” he said.

“People are still using it well, but specifically since Temple had a lot of new hires from groups that use computers in their research, and there’s also a move toward using computers in what you would call non-traditional HPC [High Performance Computing] fields, we need more capacity,” he added.

Kohlmeyer said he plans to begin finding funding sources for a new supercomputer next fall.

Liora Engel-Smith can be reached liora.engel-smith@temple.edu

Latest posts by Liora Engel-Smith (see all)

    Leave a Reply

    Your email address will not be published. Required fields are marked *

    *

    * Copy This Password *

    * Type Or Paste Password Here *

    829,513 Spam Comments Blocked so far by Spam Free Wordpress

    You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>