Are You Tired of All the Embedded Multicore Hype?

By John Blyler

Do you feel inundated with multicore coverage? Some of our readers have expressed weariness with the industry’s multicore push. Such feelings are understandable, as all of the big players have blanketed the media channels as they tout the benefits of multicore architectures. I’m not suggesting that the benefits of lower power, greater levels of performance, and competitive costs are just marketing hype. They’re not, which is why so many articles have been devoted to embedded multicore trends and technology over the last couple of years in this magazine (see referenced highlights below). Naturally, there are economic reasons why the major embedded chip companies are moving to multicore designs (see my last few Editor’s Notes). But hype or no hype, embedded multicore platforms are here to stay.

In fact, the adoption of multicore systems is a growing trend among embedded designers. According to a Venture Development Corp. (VDC) study highlighted at the recent Multicore Expo ( embedded multicore CPU systems are expected to grow dramatically from $372 million in 2007 to $2473 million in 2011. The percent of developers who are using or will be using multicore in the next 12 months is expected to increase by 55%. In 24 months, adoption will increase to nearly 79%. These are astounding predictions!

The good news is that many designers are finding that multicore architectures will solve a variety of embedded-hardware challenges. The bad news is that the software side of the embedded multicore is lacking. There is a major—and widening—gap between hardware and software capabilities in the multicore world.The VDC report noted that vendors have reported that only about 6% of their tools were ready for parallel chips in 2007. Equally troubling is the further finding that as much as 85% of all embedded programming is done in C or C++. Although these are great languages, they aren’t optimized for multicore designs. In the short term, the industry must find ways to make C/C++ more supportive of multicore architectures. The long-term solution will require a new language and set of tools.

For these reasons and others, I’ll be increasing the coverage of multicore software technology and techniques in EIS magazine. What better way to get to the crux of the software challenge than by talking directly to the leading developers of embedded multicore technology—Intel and its partners? EIS editor Ed Sperling sat down with Intel to discuss this very issue in the lead story,“Creating a Parallel Programming Language for Multicore.” Max Domeika, one of Intel’s software gurus, goes over the basics of parallelism as well. In addition, EIS contributing editor Geoffrey James takes us back to the big picture of the multicore ecosystem with an interview of Linley Gwenap—one of the most respected analysts in the microprocessor industry.

Several of this issue’s feature articles and case studies focus indirectly on both hardware and software challenges in multicore. Of course, I’ll continue to cover the embedded-hardwaredesign space as a whole—in all its breadth. For example, EIS editor Craig Szydloski looks at the growing effect of energy management or environmental factors on embedded designs in “Green Embedded Solutions Focus on Energy Management.” Other important topics include EXC form factors, the implementation of Intel’s new Atom processors, and case studies in the medical and telecom industries. Breadth as well as depth is important for today’s embedded designers.It also is critical for any publication that hopes to be of service to these talented though taxed professionals.

Pullquote (if needed):

“…Vendors have reported only about 6% of their tools were ready for parallel chips in 2007.”