what's the future for processors? will we hit a plateau and end up resorting to some new theory of machines and logic?
also, would it be possible to make a whole CPU from discreet transistors? how big would it be?
Probably some grapheme based technology.
Yeah you could build a CPU with transistors. You could build one with tubes and marbles if you wanted to (in theory), but I don't know how large either would be.
>also, would it be possible to make a whole CPU from discreet transistors? how big would it be?
With like 2,000 transistors, you could make an extremly basic CPU. Wire it up to some EDRAM, program a basic ARM VM for it, and boot linux in 18 days.
No, you would be able to pass my school's third year Comp Eng lab though, because that's what they do for a large part of it (making a CPU from scratch with discrete components).
I imagine other engineering schools do this as well.
Another unusual feature is the use of magnetic cores not only for memory but also as logic gates. These logic cores have 1, 2 or 3 input windings, a trigger (read) and an output winding. Depending on their polarity, current pulses in the input windings either magnetise the core or cancel each other out. The magnetised state of the core indicates the result of a boolean logic function. Two clock phases designated alpha and beta are used to trigger (reset to zero) alternate cores. A change from a one to a zero produces a pulse on the output winding. Cores which receive alpha trigger pulses (alpha cores) have inputs fed from gates which are triggered on the beta phase (beta cores). Transistors were expensive at the time and each logic gate requires only one to amplify the output winding pulse; however a single transistor drives the inputs of a small number of (typically 3) other cores. If more than 3 inputs are to be driven, up to two additional transistors can be driven by each core.
No, fuck x86. Fuck all of it. We need to start completely from scratch - one single 64-bit mode, none of this 16/32/64 freak show x86 puts on. Dropping x86 compatibility isn't actually a huge deal for anybody but Windows fanboys anyway - well written software should be portable to nearly any CPU architecture as long as there's a *nix on top.
We'll never hit a "plateau". We'll just reach a point where improvements in processor technology are increasingly minimal. We're already seeing it on the desktop - enterprise computer deployments are stretching from two to three and now five years and longer. There's no need compelling reason to upgrade for 99% of all use-cases.
>how big would it be?
A "standard" three terminal pnp transistor is about the size of your pinkie nail. The area of your pinkie nail now holds about a Billion transistors (give or take) on a chip coming from a modern fab. Scale up all the connecting lines.. you get the idea?