Simplicity: the key to killer flexible chip design
When minimalism burst onto the scene in the middle of the 20th century, its principles of elegant simplicity revolutionised design. Inspired by Japanese Zen Buddhism and its rejection of the unnecessary, its pared-back forms emphasised spareness – and achieved greater impact because of it.
Now, it might be an engineer’s fanciful notion, but I look at designing for flexible chips in much the same light. Far from high-spec, multi-purpose semiconductors built for heavyweight applications, these are chips with just enough performance and complexity to get the small jobs done. No more, no less. This, in combination with their flexible form factor, means they can be used in places where it would not be cost-effective – or even possible – to use silicon.
However, where silicon chips are home to billions of transistors, flexible chips currently house just thousands. So, naturally, they require a pared-back, bottom-up design approach.
But in a world where we’re accustomed to high-powered chips that address multiple use cases and still have power to spare, how do you forget established paradigms and achieve killer flexible chip design?
Back to basics
First, you need a mindset shift: we’re not starting from a point of ‘performance at any cost’ and we’re not removing functionality from an advanced chip – we’re bringing a minimalist approach to a new kind of design. So, the first thing to ask yourself is what is the bare minimum that’s required to get the task done?
It’s almost the diametric opposite of designing for silicon, where high costs and long development times dictate that your chip be suited to as many workloads as possible. And when it takes significant time and money to change your design, you need to make sure it can do anything and everything you need it to – first time.
By contrast, designing for flexible chips is altogether more agile. Each production cycle takes just a day or two – and costs significantly less – so you can afford to get a design out, test it and tweak it to hone the functionality for your unique use case.
In some ways, it’s quite satisfying – like an intellectual puzzle. How can I optimise performance and remove unnecessary circuitry to achieve a design that’s elegant, simple and hits my target power envelope?
Simplicity beyond design
Interestingly, the simplicity of flexible integrated circuits doesn’t stop at the design stage.
The production process is also pared down, omitting many of the resource-intensive stages of silicon manufacturing. There are no high-temperature processing steps and far fewer chemical steps. As a result, it uses just a fraction of the power and water of silicon production, and significantly reduces the quantity of chemicals and gases consumed.
This has the triple benefit of substantially reducing costs and minimising environmental impact, while accelerating production cycle times. In fact, a flexible IC design will typically go from tape-out to production in less than two weeks and costs around 10 times less than silicon.
Simplification is also the key to democratising chip design and lowering the barriers to entry. Reduced complexity makes it possible to optimise for individual applications very quickly, very easily and at very low cost, slashing time to market, as well as the level of upfront investment. And since getting feedback on the success of designs happens in days, rather than months, business ideas can rapidly become products, accelerating production time without the increased risk and costs associated with silicon design.
In short, the simplicity of flexible chips is their strength, and the key to a low-cost chip that is perfectly tailored to its application, produced with minimal resources and reduced environmental impact. What’s not to like?