[extropy-chat] more moore

[Spike]

> Eugen Leitl...

> Nevermind that Moore is not about performance, but 
> integration density.
> Performance (as measured by benchmarks) doesn't follow that 
> nice linear
> semilog plots. <-- How often need I say this sentence until 
> people here stop
> using Moore as sloppy shorthand for faster computers?


Im specifically interested in reducing feature size, in
some ways even more so than increasing speed.  My understanding
is that given two processors running at the same clock speed with 
similar numbers of transistors, the one with the smaller features 
would use less power.  Is that right?  For M-Brain apps, that will 
be a huge factor.  spike