The US Government plans to spend hundreds of millions of dollars over the next several years to fund the development of huge supercomputers with power beyond anything available today.
The goal is to address the most challenging problems facing science, national security and industry -- and three key computer vendors have won contracts.
Several government agencies have awarded or are about to award contracts for systems capable of sustained petascale computing speeds, which can handle quadrillions of calculations per second. To understand the scale of these planned systems, only one system on the Top500 supercomputer list released last week surpassed 100 TFLOPS.
Earlier this month, Cray signed a $200 million contract to deliver a petascale-capable system to the US Department of Energy's Oak Ridge National Laboratory by 2008. That system, based on AMD processors will be built in phases of ever-increasing speeds.
The National Science Foundation in June began seeking proposals for a larger high-performance supercomputer that could also cost up to $200 million. And this month, the Defense Advanced Research Projects Agency (DARPA) plans to award contracts valued at several hundred million dollars for two even larger supercomputers.
Planning an approach
The scale of the computing power in the new systems will be so enormous that "we have to change the way we do computational science to really take advantage of these machines," said Dimitri Kusnezov, head of the DOE's advanced simulation and computing program, which operates the most powerful supercomputer in the world today, the IBM Blue Gene/L.
That supercomputer, with more than 131,000 IBM Power processors, was easily the No. 1 system on the latest Top500 list.
"The question is, what would [scientists] do with an infinite amount of computing speed?" said Kusnezov. "What would they calculate? And I'll wager that they don't have an answer for you." Kusnezov said petascale computing levels will likely force researchers to assemble multi-disciplinary teams to task the systems with solving fundamental scientific problems.
Three vendors -- IBM, Cray and Sun Microsystems -- have been working with DARPA for several years on initial research for the next generation of computer systems. Next month, the agency is expected to pick two of the three vendors to work on the next phase of the project.
DARPA's goal is to build an "economically viable" petascale supercomputer, according to an agency spokeswoman. DARPA expects development of the system to last four years, she said.
The NSF also wants a petascale system and is seeking proposals -- due next February -- for a supercomputer that can answer key questions about the kinds of abrupt transitions that can occur in the Earth's climate and ecosystem structure. It wants the system to be ready by 2011.
Stephen Meacham, IT research programme director at the NSF, said the agency wants to use the system to attack "frontier problems," such as modelling the interaction of viruses with various components in a cell and looking for ways to block those interactions.
Although much of the focus on supercomputers is on the number of processors being strung together, more vexing problems involve memory and storage subsystems and energy consumption, which "begin to take up a good chunk of the overall cost of the system," said Dave Turek, vice president of deep computing at IBM.
The next-generation Blue Gene system, Blue Gene/P, aims to deliver petascale performance with only a 10% to 15% increase in power consumption, Turek said. IBM isn't disclosing details or shipment plans for the system.
Turek said the most important goal is building low-cost, high-performance systems that businesses can use, in order to extend the systems' use beyond government agencies.
To help support that cause, legislation was introduced in the US Senate last month by three senators to set aside $25 million to fund up to five supercomputing centres for assisting businesses and manufacturers.
The legislation, called the Blue Collar Computing and Business Assistance Act of 2006, gets its name in part from the Ohio Supercomputing Center in Columbus, whose Blue Collar Computing initiative is intended to promote supercomputing in mainstream IT.