Tackling the SOC Design/Delivery Gap

Electronic News, Nov 16, 1998 by Carolyn Whelan

edinburgh, scotland-Keeping fabs full. That's one of two driving forces behind the consumerization of electronics, according to Jack Harding, CEO of Cadence Design Systems. The other one is the wealth of intellectual property (IP) that has and will continue to be available after the Cold War's demise. IP that's led to revolutionary applications like the mouse and Internet, now ubiquitous among consumers, as well as emerging GPS (Global Positioning Systems) products.

Mr. Harding was showing his crystal ball to a room-full of Scottish dignitaries last month at a University of Edinburgh lecture, as he shed light on an innovative collaboration between companies (including Cadence), government agencies and academia-Project Alba. The joint project strives to speed up the development and take up of system-on-a- chip technology by pooling the resources of all three groups-in particular, tapping into the wealth of Scottish engineering talent to accelerate development of the new technology.

"We're talking about the advent of a single chip with 50 million or 60 million transistors on it," said Mr. Harding. "It's going to take a radically different design process." One analyst has estimates the size of the virtual component (VC) market (the building blocks of SOC) will grow from today's $17 billion to $25 billion by the year 2000, according to Cadence. If the speed with which Alba was implemented is indicative of the project's success-nine months from inception to execution-then things bode well for Alba.

"There's no such thing as a low-technology industry," added Mr. Harding. "Winners and losers will be chosen by their ability to incorporate technology into products." In this era of sensor actuators, he said, many products will incorporate features previously attributed to robots. Among new developments: retina scanners on every PC, and voice-activated everything in automobiles. "But the consumer won't pay more than $400-for anything."

Alba's a unique undertaking for several reasons. First it is tri- partite involvement, which has been unheard of until now. Also of interest are the newly established Virtual Trading Center (VTX) and the Systems Integration Institute.

So most of the elements are in place. But Alba does have one missing link: the end-market focus.

Until now, much like the driving force behind the consumerization of products, SOC design has been working in a bit of a vacuum, with technology for technology's sake. As PC prices plummet while component prices remain steady, designers have been responding to OEMs' fears about diminishing profit margins by churning out designs not necessarily based on customer needs and end-market success, but, rather, combinations of products that developers-with a little help from their OEM friends-think might fly in the marketplace. That explains why few designs have actually emerged in products; why some that have, haven't flown; and while others have emerged just plain too late. Unfortunately, Scotland has little to offer on that front.

"We've always been good at inventing things," said Andy McDonald, a manager at Locate in Scotland, a government agency. "But not so good at getting them to market." Mr. McDonald said that universities today are encouraging students to think about the commercial applications of their work. "The spin-off-it's a familiar model in the U.S., but not so in Europe."

The lack of a market focus at Alba is not surprising. Scotland has traditionally been an IT manufacturing center, churning out products defined and qualified by subsidiaries abroad. Also, academia and science play such a big role, many of who have been taught research for research's sake, with little commercial training. Cadence is the only commercial tenant on the mammoth site currently under construction, though several companies are said to be seriously considering a move there.

One reason that customer link is so critical is time-to-market. Most SOC designs, until now, have taken over two years to develop, largely due to complications associated with licensing IP to integrate onto the core. So the value-added, niche elements of the proposed product are borne early in the process rather than later, when generic components are merged in traditional electronic design. If designers miss the market, the work can be for naught.

Another reason is the new business model that SOC enables. Integrating most of the necessary components on one chip allows the designer to leapfrog over chip and component makers, straight to the OEM. It also frees designers to work directly with system integrators, who, in turn, pull the technology together for future products, which they pitch and sell to OEMs. About 75 percent of Cadence's clients are system manufacturers, and 25 percent are chipmakers. As vertical companies break up, eliminating those layers and streamlining the process puts more of an onus on the design company to know their market and get the elements right. Designers have traditionally churned out more generic designs focused on a particular industry, like telecom, television or appliances.