Re: Resistance is futile
@Wzrd1 - a fundamental misunderstanding of CMOS. In a pre-CMOS computer, a bit was represented as a flow of current. It's consuming power even if it's just maintaining an unchanging logic level for minutes on end.
in CMOS a bit of state is a package of charge - maybe as little as 100 electrons. No current flows except when a bit of state is chaged, when the electrons have to be removed from a high voltage (probably representing a 1) to a low voltage (0). CMOS can work on micro- or nano-watts. Witness the hand calculator powered by a couple of square cm of low-grade PV panel, illuminated by an energy-saving dim light bulb (and operated by a dimmer one - sorry). Easy when you want single-digit IPS not MIPS.
Moore's law is based on a scaling law. If you shrink the devices by a given factor, reduce the voltage (between 1 and 0) by the same factor, you have constant power per unit area of chip and that factor squared more devices to play with in the same area. The limiting factor is that atoms are discrete, and today we are at the point where the gates of the FETs can no longer be made much (if any) thinner. So the scaling law can't be followed any further, and the first sign of trouble is that the chip runs too hot because it's suffering resistive heating from various sources.