Sign In

Communications of the ACM

ACM News

The ­nified Theory of Wu

View as: Print Mobile App Share:
Virginia Tech Associate Professor Wu-chun Feng

One of Virginia Tech Associate Professor Wu-chun Feng's supercomputing projects harnesses the collective power of dormant desktops to tackle "some 'grand challenge' problems that have been difficult to address."

Credit: Virginia Tech

Wu-chun Feng is way too busy. An associate professor at Virginia Tech's Department of Computer Science, "Wu," as he prefers to be called, is occupied on any given day with professorial duties, massive technology and computer science education projects, and babysitting the university's resident supercomputer, HokieSpeed.

Wu has adapted to his breakneck schedule accordingly. Longtime colleagues marvel at his ability to work independently on his latest grant or writing project, all while carrying on a high-level conversation without missing a single beat. During a chat, Wu admits he's running on little sleep for days in a row — even though he recently had to cancel a talk on green supercomputing in Germany.

While Wu often finds it hard to say "No" to demands on his time, the type of high-performance computers he works on don't have to. Supercomputers like HokieSpeed and Green Destiny (another project of his) take advantage of parallel computing, or using multiple processing elements to perform tasks faster. This power — both on the part of Wu and his computers — promotes a significant cross-fertilization of high-performance computing ideas, which is probably why Wu ends up being so busy.

"That's part of the beauty of computer science and computer engineering: the understanding of abstraction," Wu says. "The way you can take some things you're working on and apply them in other ways you might not have otherwise thought of doing."

To keep all these ideas organized, Wu currently uses the Mac-based OmniFocus productivity tool to collect his thoughts, create his to-do list, and get things done. His office is almost exclusively used for administrative tasks. The real ideas come from liberal time spent outside of it — thinking, creating and, of course, playing ultimate Frisbee — to foster creativity. If only there were a way for someone to tap into Wu's creative brain when he's filing papers. While we're not there yet, he is applying the idea to computing.

Wu is part of a team of Virginia Tech researchers that are turning the university's math lab into a supercomputer by harnessing the power of ordinary computers anytime they lie dormant. Called Project Moon, the initiative could have serious commercial applications for companies that want a healthy middle ground option between using no infrastructure at all (the cloud) and entirely local computing infrastructures. Not to mention the "altruistic" aspects of the project.

"This type of supercomputing has the promise of being able to address some 'grand challenge' problems that have been difficult to address," says Wu. "Things like doing reverse engineering of the brain and finding missing gene annotations in genomes."

Wu is already pursuing those higher-minded goals by playing a big role in the Nvidia Foundation's Compute the Cure initiative, which leverages Project Moon-style desktop computer collaboration to change how cancer research is conducted.

While those efforts progress, Wu would like to see the computing world embrace parallel processors a little more vociferously.

"A lot of software out there is still written with this notion of single-threaded on a single processor," he says. "Performance improvements based on that aren't going to pan out as much anymore, because the clock frequencies on these processors have effectively come to a halt."

Reaching higher performance goals will require educating the next generation about computers, with greater attention on the practical application of computing fundamentals, Wu says. Last year, he received an IBM Faculty Award to do just that: he's using the $16,000 prize to develop a computer science curriculum for grades K through 8.

"I'd like to see education more in terms of problem solving with computers," Wu says.
"It's one thing to have the tool; it's another thing to know how to use the tool appropriately."

Going forward, the only certainty is that Wu will remain busy — very busy. But it's alright: his continuing work on parallel computing ensures that he'll have a little help from his supercomputer friends.

Logan Kugler is a freelance technology writer based in Silicon Valley. He has written for over 60 major publications.


No entries found

Sign In for Full Access
» Forgot Password? » Create an ACM Web Account