Choosing The Right Systems Design Path

I’m a cheap bastard, usually given to self-abnegation when it comes to buying material goods for myself. But I broke down and bought a runner’s watch late last year because I wanted to change up my exercise routine to run the same distances, only faster.

I quickly decided against going all in and getting a GPS watch. At this point in the arc of electronics-design technology, it’s hard to find solar-powered versions that don’t break a cheap bastard’s bank account.

So I settled on a standard solar-powered Casio model for less than $30. The question becomes, can I bide my time during the waning days of Moore’s Law until the GPS versions slip down into that range? Will it even happen? Remember, we’re all impatient electronics consumers.

Stalled Out
We seem to be sort of treading water in system design. There hasn’t been a “next big thing” in mobile systems in recent electronics shows like CES. Mobile World Congress will be the next stage and maybe we’ll see a breakthrough; maybe not. (I’ll let you know in this space, because I’ll be there at Embedded World covering the events).

The battleground is probably energy harvesting. Absent overnight advances in battery chemistry, this is the most likely way to advance the low-power/high-performance paradigm in mobile design. Or is it?

The question today is—in the context of that affordable solar-powered GPS watch—how do we get there from here? What path shall we take?

One hurdle is the power wall. Stanford professor Mark Horowitz expounded on the challenges this week at ISSCC, where he suggested climbing gear for that wall must include specialized silicon and better algorithms:

“If you think about applications that don’t access memory very much, you can see why specialization can help. Specialization is not so much about hardware, but you have to move algorithms to a much more restricted space.”

He went on to say:

“If we want algorithm designers to play and create better computing devices, we have to minimize the cost to them to do exploration. We have to give them a much higher level development platform in which to play. We can take sets of hardware and build a strong environment so designers can write code.”

Global View
Cadence Fellow Chris Rowen, in a recent presentation, echoed Horowitz’s philosophy, saying we’re entering an era of the “globalization of processors,” in which applications find their way to the most appropriate processors, and domain-specific languages take hold to ease design complexity and boost productivity.

Moving from the philosophical to the practical, Jim Hogan and Vojin Zivojnovic describe how the EDA industry needs to respond:

“Energy-efficient, complete solutions can be obtained only with optimal alignment across the pre- and post-silicon phases of energy optimization, supported by unified design flows, abstractions, and formats.”

So, back to our opening metaphor: Running to stay in shape is about changing things up, stressing your system in surprising ways to build muscles and keep fat at bay.

For decades, we ran a certain way in silicon and system design, sprinting wildly in the clock-speed race.

But you have to change things up (indeed, in our case we were forced to). Assuming no physical or lithographic barriers, we could have stayed on the clock frequency race indefinitely and gotten to that solar-powered GPS watch via that path. But that turned out not to be an option. Designers have been forced to explore and adopt new approaches such as the globalization of processors and other techniques that Rowen and Horowitz and Hogan and Zivojnovic have talked about.

And that’s OK, because you can take different running trails and end up, happily, in the same spot.