Bending the rules with flexible non-silicon 32-bit RISC-V chip Brit chipmaker Pragmatic Semiconductor has created a 32-bit microprocessor in a "flexible technology that is fully functional while flexed." The Flex-RV processor isn't about winning performance benchmarks, but creating a new bending compute solution to fit in unconventional places. That said, it does include a programmable …
The important number missing is the power requirements: 5.8mW at 3V. We have got used to power scaling with frequency but this tech dates back to the age where power was pretty much constant between full speed and stopped. This really shows up in memory where most of it sits idle while you read/write one word. To prevent power scaling with memory size this implemntation uses external memory. Instead of the massive waste of resources used for AI you should be thinking hand coded assembly language optimised to use the absolute minimum number of bits. Even with that limitation in mind this is impressive tech. The next huge problem is to get enough market share to divide non-recurring engineering costs down below silicon.
Probably a great achievement, so congragulations, but...
Really poor power consumption - this is 3-5 orders of magnitude worse that off the shelf micros. Not that that is any kind of surprise: large = big C, and C*F=power. Big will never be energy efficient.
The bigger problem is if there is any commercially significant use for it.
Something large, thin and soft is not actually very robust. By comparison I have run silicon chips with a sharp steel scriber pressing 1kg direct onto the chip surface. It is not the chips that are weak it is the transition to flex in flexible material that fails. Small, low power chips are incredible rugged. It is only really big, hot, high pin count chips that are problematic.
Hard to see any cost advantage - low performance silicon is incredibly cheap - ones of cents, certainly cheaper than a largish area of flexprint. When you add in poor power consumption, then the power source becomes bigger and more expensive.
The best use I see for this is if performance can be improved enough to be a display driver for inherently flex displays like eink on flex, polymer lcd and oled and electo??. These are inclined to have a problematic requirement for a lot of connections onto flex. While it is a solved problem, making that driver on flex might be a big cost reduce for those display techs.
A tiny silicon chip, and a flexible serial->display/lots-a-pins is a more viable outcome
Cost of manufacture versus cost to operate. Consumers have spent a LONG time not thinking about how big a load a given washing machine might be on the bills...
It wasn't so long ago that your average TV sat with half a dozen or more boxes in front of it at idle. That demand has dropped off as the mechanical needs for tape, disc, etc. have fallen away somewhat.
Interestingly, folks living off grid tend to have rather more of an understanding of watts and watt hours. My mum was clueless till she moved off grid, wind batteries and a diesel only... Got quite good at only switching on enough gadgets to avoid the diesel firing up. Otherwise, we tend not to think about it, preferring the convenience.
While reading the article my first thought was for active signage. Many moons ago the people I was working for designed these using a PC and monster CRT units. Something like this combined with E-paper would dramatically reduce running cost and bulk, while improving reliability.
And if we use colour e-ink whilst also giving the customer the opportunity to reprogram their own packs, this would once and for all resolve the age-old question, one so deep and profound that even one of Slartibartfasts finest creations would struggle with it, as to what colour a pack of cheese and onion should be in the first place...
This post has been deleted by its author
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
