When hardcore video game enthusiasts put down hundreds of dollars for a graphics card and are not constantly playing Halo or Half-Life 2, they can now help solve ailments like Alzheimer’s disease.

At least that is the hope of Stanford’s Folding@home program, a piece of downloadable software that adds a computer’s processing power to a network of hundreds of thousands of others engaged in simulating protein folding. The makers of the program have just completed an update of Folding@home to take advantage of the huge processing power in ATI Technologies graphics cards, which have been marketed primarily toward gamers.

Vijay Pande, an associate professor of chemistry and manager of the research group that created Folding@home, said that by using these cards, the group has been able to achieve speeds 40 times greater than that of the typical Pentium 4 Processor.

“That turns three years of calculation into one month,” Pande said.

Created to deal with issues too complex to run on supercomputers, distributed computer networks have been used to provide information on massive science problems. Folding@home specifically deals with the three-dimensional shapes most proteins need to fold into in order to function properly. Protein misfolding is common in many diseases, such as Alzheimer’s and Huntington’s.

Pande described the immense computational power needed to study protein folding.

“What we’re trying to understand is a biological molecule using physics,” Pande said. “Because of the application of physics, we don’t need to rely on prior experiments, but work from first principles. But since atoms move extremely fast, the time increments we need to study are very short, around a quadrillionth of a second.”

There are problems, however, with using just one computer to map protein formations.

“It’s possible to do around one million iterations in a day,” he said. “So that gets you up to one billionth of a second. It would take one billion days on one computer to get even one second of data. Now, not much that’s interesting happens in a billionth of a second, but some interesting things start to happen around one millisecond. But even that is one million days, which come out to roughly thirty thousand years. But if you have one hundred thousand computers working on the problem at the same time, that only takes ten days.”

Folding@home was released six years ago in late 2000. In order to make the program function properly, the team had to develop the algorithms needed to enable hundreds of thousands of computers in distant locations to work in concert.

“Those algorithms are very important,” Pande said. “You could imagine having 120 students taking an exam in an hour. The professor says to the students that they can all work together on the exam, but they have to finish in 30 seconds.”

The algorithms also minimize demands on individual users’ Internet connections. Typically a computer will be assigned a calculation that takes anywhere from a few hours to a couple days. The result of the calculation is then relayed to a central server. At most, only a couple megabytes of information is passed to the client computer and back, meaning that even users with slow connections can participate.

Folding@home is currently running on 200,000 computers and, in the six years since its release, nearly two million computers have participated. As a result, the Pande Group has published 45 scientific papers using Folding@home’s results.

After developing technology for ATI graphic cards, Folding@home is now working to expand to video game consoles.

“We’re also leasing software for the PS3,” Pande said. “Apparently the chip in the PS3 is very fast. Graphics Processing Units and the Playstation 3 use something called a streaming processor. Those streaming processors do very well for games, the new thing here is our ability to use them for science.”