8800 GTX gonna pwn you in September

[tarna]

Are there problems with the length of wire...

Wire length is a HUGE problem in general. As clock speeds keep increasing, one of the single largest problems chip designers have to face today is the amount of time it takes for a signal to travel from one end of a chip to another. It is very hard to do this in one cycle nowadays. Transistors are becoming smaller and faster, but wire delays are becoming increasingly dominant.

 

<{POST_SNAPBACK}>

There is also the problem of 'cross-talk' between the wires. As the frequency of a signal increases, so does it's likelihood to 'broadcast' it's signal to the next wire to it causing all the wires to read the same signal. I built a calculator in college that had this problem because I used wires instead of a printed circuit board. It was a better conversation piece than it was a calculator.

Right now, the frequencies used are still in the radio frequency range. I'm looking forward to see what happens when they hit frequencies usually associated with light. The problem may go away because of insulation or they may become something else entirely.

[angshuman]

There is also the problem of 'cross-talk' between the wires. As the frequency of a signal increases, so does it's likelihood to 'broadcast' it's signal to the next wire to it causing all the wires to read the same signal. I built a calculator in college that had this problem because I used wires instead of a printed circuit board. It was a better conversation piece than it was a calculator. 

Right now, the frequencies used are still in the radio frequency range. I'm looking forward to see what happens when they hit frequencies usually associated with light. The problem may go away because of insulation or they may become something else entirely.

And don't forget those damn atmospheric neutrons!

 

EDIT: I just noticed you live in Denver. :D

Edited August 2, 2006 by angshuman

[tarna]

Can't believe that I missed this for over a week. ( grumble, grumble )

Magnetics obviously isn't something to be ignored when it becomes an influence from wire to wire. That's why printed circuits became the staple of the integrated circuit industry. 'Cross-talk' is the old-fashioned term for magnetic interferance across lines causing all signals to read the same via hysterysis ( sp? ). All digital signals = 1.

 

I'm sure you were teasing me but I had to respond.

[angshuman]

I think you misread me . My comment on atmospheric neutrons wasn't meant to be an attempt to downplay the existence or seriousness of the crosstalk issue... those damn neutrons are seriously becoming a pain (just like crosstalk is). I mentioned Denver because the neutron flux there is among the highest in the continental US (you'll see the city being mentioned in a lot of literature on neutron-induced bit-flips).

[tarna]

:"> Since I haven't been into electronics for a while I hadn't heard anything about that and thought you were teasing me. ( sets aside tin foil hat )

What impact does neutron flux have on circuits? I would have assumed that since a neutron has no charge that it's effect could be ignored.