August 11, 2010

DARPA Funds Nvidia Quest For Supercomputer Chip

Nvidia announced Monday that it will be leading a team awarded a 25-million-dollar contract by Defense Advanced Research Projects Agency (DARPA) to create a microchip that it hopes will "leave today's supercomputers in the dust," accfording to a recent AFP report.

DARPA wants to make supercomputers as much as a thousand times more powerful with technology from graphics processing units (GPUs) used to power realistic video gaming action.

GPUs break complex tasks into separate parts and handle them simultaneously while central processing units typically used in computers tend to projects in sequence, rushing from start to finish in order.

The Nvidia team includes Oak Ridge National Laboratory in Tennessee, supercomputer company Cray of Seattle, and a group of six US universities.

Intel will also be at the forefront of a team that will also work toward building a faster chip, but the company has not detailed who will be on its team or what its award will be as of yet.

Nvidia and Oak Ridge are already working on hybrid CPU-GPU computer clusters, and last October announced the US Dept. of Energy had kicked in funds to study the use of Nvidia Tesla GPUs in x64 clusters.

Oak Ridge is also where the 1.76 petaflops Jaguar Cray Opteron-Linux cluster is housed as well. So, in reality, DARPA is paying for work that the Nvidia team was already pursuing piecemeal, but shaping it for a particular physical environment and code set.

"This recognizes Nvidia's substantial investments in the field of parallel processing and highlights GPU computing's position as one of the most promising paths to exascale computing," said Nvidia chief scientist Bill Dally.

"We look forward to collaborating to develop programmable, scalable systems that operate in tight power budgets and deliver increases in performances that are many orders of magnitude above today's systems," he said.


On the Net: