Mike Henderson knows trucks. He’s president of a company called Smart Truck, and he had a problem.
"Trucks," he says, "are fairly unaerodynamic devices. Half their fuel goes to defeating aerodynamic drag.”
But he needed to make his clients’ semi-trailers more fuel efficient to comply with California law -- and fast. Trouble was, calculating airflow is crazy complicated.
There are all kinds of vortices. "Almost every piece of a truck produces wake and they mix with other wakes," Henderson says. "It would take over a week to do a single calculation" to see how a single part would interact with a whole truck.
And he’d need to do hundreds of calculations.
So Henderson did what a growing number of businesses are doing: He managed to get access to a supercomputer. Not just any supercomputer, but "the most powerful computer in the world for open science," according to Suzy Tichenor at the Department of Energy's Oak Ridge National Laboratory.
Titan, as the $100 million supercomputer is known, is nearly the size of a football field.
"It clocks in at about 27 petaflops, which is the equivalent of 27 quadrillion calculations or operations a second," Tichenor says. A quadrillion is a million billion. "It’s a big boy."
Labs and businesses – even small businesses like Henderson’s -- can apply to use it. It’s free if they make some of their results public, thereby fulfilling Oak Ridge’s goal of promoting advancement in science and engineering. Henderson was able to test prototype trucks that existed only inside the computer before going to the trouble of making real parts. His team saw air movements that would’ve been invisible in a wind tunnel.
"It probably got us to market a year faster than we would have had we not used it," he says, adding that he improved trucks’ fuel efficiency by 14 percent by adding special airfoils.
Tichenor says other businesses have been banging down the door for a crack at Titan. The number of applicants has risen 20 percent in the past year.
"We’ve definitely seen a growth in the number of companies that are applying because modeling and simulation allows you to accelerate the research and development process dramatically," she says.
That includes auto companies which would rather not build a hundred different cars just to crash them all, insurance companies modeling how fires spread and biotech companies too.
Not only have firms been applying to use government supercomputers, they’ve been buying up their own.
"The market has been growing at a rate we’ve never seen before," says Earl Joseph with international market research company IDC. The company has been tracking the computer industry for two decades. Joseph says a medium size supercomputer costs $10 million, but they’ve been selling like hotcakes even through the recession.
"Supercomputers grew 65 percent in 2009 alone, last year sales grew another 29 percent. You’re not seeing those growths in any other sector right now," he says.
There are a few reasons supercomputers are now a $5.5 billion industry. "The questions, the problems that users are looking to solve are becoming more and more complex," says Barry Bolding at supercomputer producer Cray, which made Titan. Demand for Cray’s computers has increased five-fold over the past five years.
BP, for example, has an array of seismic sonar devices the size of Manhattan. Processing the echoing sound waves to suss out the contours of oil and gas deposits beneath the ocean floor is an immense computational task. Another factor: Computers have become faster by incorporating processors known as GPUs that were developed for the gaming industry.
Earl Joseph, at IDC, says the next milestone in supercomputing will be something called an "exascale machine," a thousand times faster than Titan.
"The biggest problem right now is the electrical bill," he says. "To build that big of a computer you almost need a nuclear power plant right next to it." The annual electricity bill could run anywhere from $30 million to $50 million.
He says someone will probably figure a way around that though, as governments and companies alike run a supercharged race to build computers that are ever more super.