In a squat cement building at the outskirts of Virginia Tech’s campus, 1,100 Macintosh computers are stacked like library books–arranged by students in exchange for football tickets and pizza.
Despite its humble appearance, this cluster of off-the-shelf G5 Power Macs, or the “Big Mac” as students have begun calling it, is about to rank as the world’s third-fastest supercomputer, at 10.3 trillion operations per second.
The cost: About $7 million, significantly less than the custom supercomputers that labs use for weather and weapons simulations, chemical experiments, and other highly complex projects.
“It’s really quite impressive,” said Jack Dongarra, a computer science professor at the University of Tennessee who compiles an annual list of the top 500 supercomputers. “They’re now competing in terms of performance with our foremost research facilities.”
Dongarra, who was about to release the 2003 rankings on Nov. 17, said he expects the only faster supercomputers to be the Earth Simulator Center in Japan, which cost at least $250 million and has been clocked at 33.9 trillion operations per second, and a computer at Los Alamos National Laboratory (clocked at 13.9 trillion operations per second) made for $215 million.
As impressive as Big Mac is, some experts believe the supercomputer business probably need not fear a new generation of home-brew rivals.
“To be honest, these clusters have their trouble points,” said Ed Seidel, who supervises a computer cluster at Louisiana State University. “You know how often your own PC fails sometimes? Just think of 1,000 computers. Custom-built systems are more reliable at times–they were engineered to be supercomputers.”
Even so, Virginia Tech’s computer cluster dwarfs the power of other computer clusters.
Each of the 1,100 Macs has two IBM PowerPC 970 microprocessors that are based on a relatively new 64-bit design, which means they process data in chunks of 64 bits–a method exponentially faster than older, 32-bit technology. The processors are connected by a high-speed network called Infiniband that allows them to break up major calculations and analyze each part at the same time.
Virginia Tech quantum chemist Daniel Crawford said Big Mac will shorten the time he spends building computer models of chemicals from years to just days.
“It opens an entire new area of chemistry,” said Crawford, who will be one of the first to use the computer in December. Before Big Mac, Crawford would have had to wait in line to use part of another computer. “You could spend your entire career on just a few calculations,” he said.
Project leader Srinidhi Varadarajan said other universities likely will follow with even faster clusters, drawn as Virginia Tech was to the cheap components and the pressure to enhance their academic reputation despite severe budget cuts.
Dan Powers, vice president of grid computing strategy at IBM Corp., says he’s not worried about such off-the-shelf PC clusters threatening sales of traditional supercomputers.
For one thing, Powers says, the Big Mac project took a lot of labor that Virginia Tech was able to get for free.
Virginia Tech’s 160 student volunteers spent days removing the computers from their boxes, installing communications software, and loading them on rows of liquid-cooled metal racks.
Powers said almost all supercomputer users would rather pay the $100,000 to $10 million for a supercomputer than try to build their own. (Highly advanced models range closer to $200 million.)
But Apple Computer Vice President Greg Joswiak said the company is getting phone calls from other institutions interested in duplicating Big Mac’s success.
Varadarajan and his staff have spent the past six weeks running test programs, teaching Big Mac to think faster. The supercomputer already has added about 3 trillion operations per second to its initial performance benchmark.
“This is like tweaking a race car,” he said.
Va. Tech’s Terascale Cluster project
Dongarra’s “Top 500 Supercomputers”