DARPA has an ambitious $1.5 billion plan to reinvent electronics

[Editors Note: See also Cadence Selected for DARPA ERI Machine Learning Contract]

Hardware innovation has taken something of a back seat to software advances in recent years, and that bothers the US military for several reasons.

There are also worries about the rising cost of designing integrated circuits, and about increased foreign—for which read “Chinese”—investment in semiconductor design and manufacturing (see “China wants to make the chips that will add AI to any gadget”).

The ERI’s budget represents around a fourfold increase in DARPA’s typical annual spending on hardware. Initial projects reflect the initiative’s three broad areas of focus: chip design, architecture, and materials and integration.

One project aims to radically reduce the time it takes to create a new chip design, from years or months to just a day, by automating the process with machine learning and other tools so that even relatively inexperienced users can create high-quality designs.

“No one yet knows how to get a new chip design completed in 24 hours safely without human intervention,” says Andrew Kahng of the University of California, San Diego, who’s leading one of the teams involved. “This is a fundamentally new approach we’re developing.”

“We’re trying to engineer the craft brewing revolution in electronics,” says William Chappell, the head of the DARPA office that manages the ERI program. The agency hopes that the automated design tools will inspire smaller companies without the resources of giant chip makers, just as specialized brewers in the US have innovated alongside the beer industry’s giants.

New chip materials and clever designs

If we’re going to move beyond Moore’s Law, though, the chances are that radically new materials, and new ways of integrating computing power and memory, will be needed. Shifting data between memory components that store it and processors that act on it sucks up energy and creates one of the biggest hurdles to boosting processing power.

Another ERI project will explore ways in which novel circuit integration schemes can eliminate, or at least greatly reduce, the need to shift data around. The ultimate goal is to effectively embed computing power in memory, which could lead to dramatic increases in performance.

On the chip architecture front, DARPA wants to create hardware and software that can be reconfigured in real time to handle more general tasks or specialized ones such as specific artificial-intelligence applications. Today, multiple chips are needed, driving up complexity and cost.

Some of DARPA’s efforts overlap with areas already being worked on extensively in industry. An example is a project to develop 3-D system-on-chip technology, which aims to extend Moore’s Law by using new materials such as carbon nanotubes, and smarter ways of stacking and partitioning electronic circuits. Chappell acknowledges the overlap, but he says the agency’s own work is “probably the biggest effort to make [the approach] real.”

Read more..