“Selling System-Level Design”

By Frank Schirrmeister
Between reviewing what happened in 2011, trying to predict what 2012 will have in store, and planning activity for the system-level design product line I am working on at Cadence, I ran across my notes and the summary of Jeff Cox’s book “Selling the Wheel”.

As Silicon Valley high tech marketers we all have been accustomed to Geoffrey Moore’s “Crossing the Chasm” and its successors to be our main textbooks.  “Selling the Wheel” made me take a fresh look at the state of system-level design. It sends a message consistent with “Crossing the Chasm”, but puts its lessons into a different time, the stone age. Its main hero Max is a resourceful fellow who invented the wheel and found himself faced with the challenge of convincing people to accept his breakthrough innovation. While I followed Max and his wheel through the four distinct stages of technology adoption – birth, fast growth, incremental growth and maturity – I couldn’t help but think about system-level design and which state its technologies are in.

In the first stage – “birth” – the technology is described as new and revolutionary, yet primitive. Products are capable of only basic tasks and have few, if any, features or options. During this period of infancy, appeal and value are limited to a relatively small number of people – but, if the technology is successful, those few quickly become enthusiasts.

In the second stage of “fast growth”, the technology advances, often dramatically and in big jumps. These advancements increase options and complexity. The implementation of the technology is often custom tailored to each buyer’s special needs. Though now more valuable to a broader range of people, the technology still has many skeptics.

The third stage sees “incremental growth”. Now the technology goes into widespread use. Market penetration increases and the technology becomes accepted by the majority of the potential market. Though the technology still continues to advance, improvements come in smaller, less dramatic steps. The products themselves become feature rich, and options become a marketing tool. Those who buy and use the products are familiar with performance issues and can hold informed conversations about them, often to the irritation of the dwindling number of holdouts.

Finally, in its last stage, technology reaches “maturity”, is standardized and has near-universal acceptance within the culture. Users are considered eccentric if they do not use it. Advancements are few and might be met with resistance. Choices of features and options diminish as they become standard to the essential product. The products itself are simple to use, even “fool-proof”. But this causes the technology to become frozen, thereby creating opportunities for the next new technical wave that will force its obsolescence.

So how do System-Level Design technologies fare in this classification?

Let’s use the EDA Consortium definition for “Electronic System-Level Design, Synthesis and Verification, which separates the portions “Design”, “ESL Synthesis”, “Verification” and “Virtual prototyping”. Design has the sub-categories “Algorithm Design and ESL Modeling”, “Design Assembly”, “System-level Architecture Analysis”. This captures pretty well the technologies the industry invented (repeatedly) over time to deal with abstracting away design complexity as shown in the graph below.

Abstraction Progression to Transactions and EDAC ESL Categories

In thinking about those technologies, I would classify them as follows:

• “Algorithm Design” is definitely in the “Maturity” stage, with tools like Matlab, Simulink and SPW being available for at least two decades. They may not be connected the rest of the hardware design flow as of yet as well as they could, but in itself they are certainly mature.
• The amount of skepticism I hear daily leads me to believe that “Virtual Prototyping” is in the “Fast Growth” stage. This goes also well with all the buzz around this technology.
• “Architecture Analysis” has been around for almost as long as “Algorithm Design” with tools like Ptolemy and various commercial offerings. It is probably either in “Incremental Growth” or even “Maturity”. Its market is just pretty limited because the number of architects per project is very small.
• Assuming “ESL Verification” includes transaction based verification, this area most definitely is at the verge of maturity. We are looking at feature rich methodologies, but also users not applying them are considered eccentric, a characteristic of the “maturity” stage.
• “ESL Synthesis” probably also is in the “Incremental Growth” stage, with some customers having reached more than 50 tape outs involving ESL synthesis. Given that ESL synthesis complements “IP Re-use” and with that “Design Assembly”, it may even be considered in the “Maturity” stage.
• Are there any ESL technologies in the “Birth” stage? Automated “Design Assembly” at the TLM level probably fits here – most of the assembly is still done at the RT level. Other technologies in the “Birth” stage would be technologies like UML and SysML in their application to HW/SW design. I have seen early customers using them as the next design entry above TLM, but their number is extremely small. Bringing together software and hardware, using embedded software for hardware verification is probably still in this stage, albeit getting closer to “Fast Growth.”

Having looked at ESL from this perspective, it is surprising how many technologies are beyond their initial stages of adoption. It may just take better connections between the individual ESL technologies as well as connections into the implementation flows, to fully capitalize on selling the wheel, um, System-Level Design.

–Frank Schirrmeister is group director for product marketing of the system development suite at Cadence.