Energy Department Pursues Exascale Computing

Military Transformers: 20 Innovative Defense Technologies

The Department of Energy has awarded contracts to AMD, Intel, Nvidia, and other technology vendors as part of its long-term plan to develop extreme-scale systems that operate at one quintillion floating point calculations per second.

Such systems would be 1,000 times faster than a 1-petaflop supercomputer. As a point of comparison, the world's fastest supercomputer, Sequoia, runs at 16.3 petaflops.

Under its extreme-scale research and development program, called FastForward, the Energy Department awarded contracts in three areas--processors, memory, and storage and I/O. FastForward is funded by the department's Office of Science and the National Nuclear Security Administration and is contracted through Lawrence Livermore National Security, on behalf of seven national laboratories (Argonne, Lawrence Berkeley, Lawrence Livermore, Los Alamos, Oak Ridge, Pacific Northwest, and Sandia).

"The challenge is to deliver 1,000 times the performance of today's petascale computers with only a fraction more of the system's energy consumption and space requirements," said William Harrod, division director of research in the Department of Energy's Office of Science's Advanced Scientific Computing Research, in a statement.

[ See Apple: Exiting EPEAT Environmental Program Was Mistake. ]

Intel Federal received two subcontracts, worth a combined $19 million, to work on more energy-efficient processors and next-generation memory architectures. AMD was awarded a $9.6 million contract for processor research and a $3 million contact for memory research.

A team of companies led by Whamcloud, including the HDF Group, Cray, and EMC, received a subcontract to conduct R&D on I/O and storage. Whamcloud said components developed in the project will be open sourced.

Nvidia received a $12.4 million award for research into processor architecture, circuits, memory architecture, high-speed signaling, and programming models for exascale computing at "reasonable" power levels.

"One of the great challenges in developing such systems is in making them energy efficient. Theoretically, an exascale system could be built with x86 processors today, but it would require as much as 2 gigawatts of power--the entire output of the Hoover Dam," Bill Dally, chief scientist and senior VP of Nvidia Research, wrote in a blog post. "The DOE's goal is to facilitate the development of exascale systems that consume less than 20 megawatts by the end of the decade."

Nvidia's work will build on research started two years ago on behalf of the Defense Advanced Research Projects Agency, Dally added.

The Office of Management and Budget demands that federal agencies tap into a more efficient IT delivery model. The new Shared Services Mandate issue of InformationWeek Government explains how they're doing it. Also in this issue: Uncle Sam should develop an IT savings dashboard that shows the returns on its multibillion-dollar IT investment. (Free registration required.)