I recently wrote about ARM not having the huge presence in the embedded space. I stand by my comments. But after hearing some of the rebuttal, it’s clear that the biggest difference I have with some of the folks who disagree with my viewpoint (you mean I’m not always right?) is more in the definition of an embedded system than in ARM’s stake in the embedded space.
So let’s move to a space that I’m sure we can agree on (he says with a high level of confidence). ARM is clearly very good at low-power design. Again, I can point to the huge (ridiculous?) market share ARM has in the mobile phone space. You don’t get designed into that many mobile devices without having a clear understanding of how to implement low-power design.
Where you could argue this point – and this time I’d agree – is that low power is far more about system design than chip design. But we have to start somewhere. And if you don’t build the hooks into the processor as ARM has done, it’s far more difficult to reduce overall system power.
In most cases, lowering the power consumption in your system comes with some tradeoffs. Depending on your end application and lots of other factors, you hopefully can live with those tradeoffs and still produce a highly marketable product. It depends on how long you need the system to live; how much size and weight you can afford; the bill of materials (BOM) that’s acceptable; and so on.
To really get a handle on all these aspects, and lots more, you might want to take in the ARM TechCon, which is taking place next week in Santa Clara, CA. There’s a whole track devoted to low-power design, and that includes discussions at both the chip level and the system level. Of particular interest is a talk by Jack Ganssle, who is always entertaining. He’ll be delivering a talk called Marketing Malarkey and the Truth About Low-Power Design. There’s also a talk on solar-powered design, and another on sub-volt IoT designs.
I’ll be taking in some of these sessions, and then I’ll come back to the debate of “what is low-power?”