Design Thinking for Electronic Designers--Pros and Cons of Divide and Conquer

Divide and conquer is an effective strategy for solving complex problems. Essentially, it breaks down a seemingly insurmountable problem into multiple smaller, more solvable parts. Once these parts are solved individually, the results are combined to obtain the complete solution.

One of the greatest practitioners of divide and rule was the ancient Roman Empire. They successfully used a method they called "Divide and rule" to amass and manage a vast empire that spanned three continents. They were so good at it that it became an ingrained and unquestioned part of the daily activities and governance of the empire. In fact, many historians believe that Rome's blind faith in the need to divide and isolate the various groups it ruled was a major factor in the stagnation and eventual decline of the empire. As one Roman senator said at the time, "We spend so much energy dividing that we sometimes forget to rule!"

Electronic designers have essentially used a divide-and-conquer approach to the increasingly complex task of developing electronic devices and products. We divide the design problem into hardware, programmable hardware, and software components, and focus on each bit almost independently. At some point, these components are combined to form the final product.

While this approach to design has served us well in the past, I think we are facing a Roman crossroads in how we develop electronic products. Being obsessed with a series of isolated design processes can cause us to lose sight of the wider possibilities.

Progress is not a linear process, it is prone to jarring jumps and discontinuities, the last discontinuity that affected electronic design was the widespread availability of microprocessors. This fact allowed us to fundamentally change the way we approach electronic design problems and seek software solutions to complex hardware challenges. It also allowed us to explore electronic product design methods that were previously unavailable. Essentially, software is the first major area in electronic design. Shifting some of the complexity of the hardware domain to the software domain, solving problems in both domains, and then combining the hardware and software to form the final product is a classic divide and conquer approach!

Programmable digital hardware in the form of FGPAs has been around for a long time. Its design is another "island" in the development process. Perhaps for this reason, its use has so far been mostly limited to replacing fixed hardware logic blocks on boards. But perhaps we don't see the greater possibilities because we don't connect the breakpoints - breakpoints that are spread across all design domains rather than scattered within specific design domains.

As Rome discovered (too late), the divide-and-conquer strategy has some unintended consequences. One is that it hinders the exchange of ideas between cultures. Although each culture will move forward, it is extremely difficult to unite all the culturally disparate groups in the same direction. The result is social stagnation.

Programmable logic devices have now reached a performance/capacity/price "inflection point." Like the microprocessor before it, this technology offers the tantalizing prospect of a quantum leap in our electronic design capabilities. We can begin to develop hardware as easily as we develop software, we can move functionality between hardware and software at will, and we can update hardware with virtually no limits.

However, unlike the microprocessor revolution, this discontinuous leap forward requires that we no longer segment the design process to "govern" the field, but rather unify our current design scope in order to move forward as a whole.

If we want to use programmable devices as filters to blur the lines between hardware and software, then we need to design in a way that allows us to have natural cooperation between all the components that make up the final product. We need to be able to push functionality between domains at will and have all the components keep in sync as we do.

In short, we have an opportunity to raise the level of abstraction across our entire design process. I’m not talking about incremental improvements to individual pieces of the design puzzle. Simply making software development or hardware design more efficient isn’t enough. We need to tear down the walls that stand between the various design streams and take our overall design strategy to the next level. This will open up a whole new world of design.

To do this, designers and design tool companies alike need to take a long-term view, abandon their preconceptions of how electronic design should be done, and start looking at how we can bring our communities together as one. It’s time to move beyond the silos of the design process and come together to overcome the challenges of creating the next generation of smart, connected electronics.