Rethinking Multi-Processing System Design and Verification

In 2006, software programmers will realize that their existing tools do not work effectively for the new class of multi-processor chips. While this prediction may seem to some like an oversimplification of an emerging challenge, I guarantee that it's about to have a profound effect on the electronics industry. And, it will be an opportunity for those willing to seize it.

The industry is moving from the uni-processor approach to multi-processor design, driven by power and clock speed limitations and software-enabled product flexibility. As a result, the use of multi-processor based ICs (MP ICs) is rapidly increasing. This trend is driving the critical need for a new methodology that encompasses the various hardware, software, architecture, and verification specializations. If designers can apply functionality across a range of processors, successfully debug and validate the software and hardware simultaneously, this new design paradigm will yield significant benefits for all.

Consider the popular consumer electronics product, the DVD player. It now includes five or so processors, offering tremendous functionality to the consumer. As you might expect, these designs require much more sophisticated software. Software programmers can no longer wait for the chip to debug the software because the chip is the software. The software programmer mentality that a C compiler and free GCC and GDB tools works fine for a single processor design just doesn't work for multi-processor design.

By contrast, hardware designers have had an arsenal of electronic design automation (EDA) tools for many years. That's because they know what they need to get the design right. Their methodology starts with a specification in a design language and includes formal verification and simulation for showing design correctness, and then synthesis tools to get from these high level (RTL) designs to implement the design. Hardware designers, of course, also have sophisticated hardware and software debug tools. The average price per seat for these tools is approximately $150,000, as compared with only a few thousand dollars for software programming tools.

Providing a new methodology is a significant challenge, particularly with the need to cross traditional industry boundaries between EDA, embedded software, and other areas, while touching many established design flows. Fortunately there are solutions on the way, creating industry discontinuities while producing a technological revolution along the way.

2006 promises to be the year where many smart people attempt to solve the next challenge—designing faster, cheaper, more power efficient and reliable chips that drive consumer electronics—ushering in exciting changes in electronics system design.

Multi-processing systems require new thinking about design and verification across multiple specializations on a scale not witnessed before by the traditional EDA and embedded software industries. This certainly means a discontinuity in the electronics industry, but it creates tremendous opportunity. Are you ready?