Northrop Grumman nears increasing embedded computing performance per Watt to as high as 75 GFLOPS

ARLINGTON, Va. -- U.S. military researchers may be just a year away from developing embedded computing technology that boosts performance per Watt dramatically without increasing size, weight, and waste heat.

In fact, researchers may be on the verge of increasing embedded computing efficiency from about 1 billion floating point operations per second per Watt (GFLOPS per Watt) to as much as 75 GFLOPS per Watt.

Officials of the U.S. Defense Advanced Research Projects Agency (DARPA) in Arlington, Va., announced a $9 million contract modification Tuesday to the Northrop Grumman Corp. Aerospace Systems segment in Redondo Beach, Calif., for the third and final phase of the Power Efficiency Revolution For Embedded Computing Technologies (PERFECT) program.

This project is taking a revolutionary approach to processing power efficiency, which includes near-threshold voltage operation and massive heterogeneous processing concurrency, combined with using the resulting concurrency and tolerating the resulting increased rate of soft errors.

Tuesday's contract modification to Northrop Grumman increases the total value of the company's DARPA PERFECT work to $31.5 million.

Related: The future of high-performance embedded computing

DARPA launched the PERFECT program in 2012. Other contractors who have been involved in the program include Columbia University in New York; the IBM Thomas J. Watson Research Center in Yorktown Heights, N.Y.; BAE Systems Electronic Systems in Burlington, Mass.; and Reservoir Labs Inc. in New York.

The PERFECT program will capitalize on expected industry fabrication geometry advances to seven nanometers, DARPA officials say. The project is dealing directly with some of the primary limitations to advancing military embedded computing: power, thermal management, and size and weight.

Current embedded computing systems have power efficiencies of around 1 GFLOPS per Watt, yet DARPA officials say current needs call for at least 50 GFLOPS per Watt, and at least 75 GFLOPS per Watt will be necessary in the near future.

In the past, embedded computing technology could rely on increasing computing performance with each processor generation per Moore's Law, which anticipated doubling the number of transistors in each new generation, with clock speeds increasing by about 40 percent for each new generation without increasing power density. This allowed for increasing performance without the penalty of increased power.

However, this free ride in processing performance increases is over, DARPA scientists declare, because increasing clock speeds now results in unacceptably large power increases.

Related: Are we in the midst of an embedded computing industry shakeout?

In the PERFECT program's first phase, Northrop Grumman developed concepts that showed enough promise to justify additional research. In the second phase, Northrop Grumman developed technology and techniques to improve overall processing power efficiency by 75 times.

Now Northrop Grumman moves to the third phase, which will provide the path to full system designs able to achieve 75 GFLOPS of computing power per Watt. By the end of this phase in January 2018, DARPA will make a set of fully mature technologies and techniques available to the Pentagon and its suppliers to implement 75 GFLOPS per Watt as products or as open-source offerings, DARPA officials say.

The program also will offer a simulation environment with PERFECT program-developed technologies and techniques to enable system design development.

On this contract modification Northrop Grumman will do the work in Redondo Beach, Calif.; Manassas, Va.; Livermore, Calif.; Cincinnati; and Littleton, Colo., and should be finished by January 2018.

For more information contact Northrop Grumman Aerospace Systems online at www.northropgrumman.com, or DARPA at www.darpa.mil.

Learn more: search the Aerospace & Defense Buyer's Guide for companies, new products, press releases, and videos