If you cooled a cpu to absolute zero, would it run at any speed?

[zsouthboy]

Has been going through my mind for some time now.

Absolute zero would leave heat out of the equation for cpu speed, no?

Or is there still a limit because the gates have a point at which they just can't switch any faster?

 

[white]

you'd probably freeze out the carriers and then your cpu wouldn't work.

 

[zsouthboy]

assuming that you somehow managed to get around that?

this is theoretical anyway

 

[spidey07]

Heck I would think at 0 Kelving you would be just about super conductive?

Theorectically I think NO, you could not run at any speed. There is still a time element involved in switching the gates any any propogation delay in the transistors.

-edit- disregard any typos, you know what I mean.

 

[imgod2u]

The only way it would be 0 Kelvin would be if the electrons weren't moving at all. Even assuming that you were in an environment in which absolute zero was possible (no background microwave radiation), in order to propogate the electrons (i.e. block them and accelerate them), you would need some type of energy and that energy directly translates into heat once it's used.

 

[f95toli]

With the cooloing technology we have today we can cool a chip down to around 20 mK (0.020 K above absoulte zero), the "bad news" is that Si-technology does not work at this temperature but you can actually get around if you use GaAs or GaN technology instead (direct bandgap devices).
However, from a "processor point of view" nothing really changes, the material is still resistive (impurity scattering) and you can not really run at higher speeeds since this is ultimatly limited by the bandgap of the material, a propertie which is not temperature dependent.

So the answer is unfortunately no.

Note:
GaN and GaAs are acutally "faster" than Si but that is true at any temperature, that is why GaAs HEMTs (High eletron mobilty transistors) are used in cell-phones.

 

[white]

yeah, at 0K just about everything becomes superconductive. so your gates wouldn't work properly.

edit: also, band gap is temperature dependent. as temperature increases, the crystal expands and the bonds are weakened. weaker bonds require less energy to break a bond and get an electron in the conduction band. this results in a lower band gap. at room temperature the band gap of Si is 1.12 eV but at 0K, it is 1.17 eV.

 

[Shalmanese]

At 0K, entrophy dissapears so you cant do any meaningful computation.

 

[Coldfusion]

No. Absolute Zero is the absense of heat (read: energy). Electricity itself theoretically can not exist at absolute zero.

 

[Fallen Kell]

When I read the title I came in to say that "No, it would not work as electrons would not be moving at that temperature, thus since there is no electron movement, there can be no electron flow, (i.e. electricity) and without electrons to carry the information bits, there can be no processing of information." Follow that to the conclusion that computing using electrons would not be possible at that temperature, so that mean CPU's as we know them today (and in the past and in theory) would not work at all in this environment. But like I said, others seem to have beaten me to this answer...

 

[Smilin]

I'd like to step in here and put a little detour on this thread if I may:

Lets assume he means "close" to absolute zero....otherwise this discussion is going to get all metaphysical and words like entropy and whatnot are going to get thrown around.

 

[zsouthboy]

Ah. I didn't know absolute zero meant abscence of energy completely.... i see now ..

 

[crzylgs]

0 Kelvin means the electrons arent moving like they said above...so yea your proc would not run at all heh.

 

[gururu]

absolute zero implies the absence of all molecular motion.

any form of energy applied to to the mass however will immediately result in the absorption of energy by the molecular bonds of the mass leading to the conversion of applied energy into heat. The only way to have energy in an absolute zero environment is to completely remove mass from the environment and the only way to have mass in an absolute zero environment is to completely remove energy from that environment.

edit: absolute zero can be maintained when molecules 'stop'. subatomic particle movement around the
nucleus is maintained and does not generate heat. that is, electrons skipping across nuclei doesn't generate heat. its electrons that crash into molecular bond orbitals that generate heat. if a material had perfect zero resistance, I would conclude that electrons could be channeled through a medium at absolute zero. of course, this is not possible, as even superconductors have some resistance.

 

[f95toli]

Originally posted by: Smilin
I'd like to step in here and put a little detour on this thread if I may: Lets assume he means "close" to absolute zero....otherwise this discussion is going to get all metaphysical and words like entropy and whatnot are going to get thrown around.

I would say 0.02 K (0.017 K in a good cryostat) is quite close to absolute zero. We have been able to do that for a long time, nothing really changes.
In fact, it is rare to see drastic changes in material properties such as conductivity once you get below about 20K because other, non-thermal, processes starts to dominate. On exception is of course superconductivity but I can not think of one single material that has a superconducting transition temperature below 20 mK at normal pressures. The main reason why it is interesting to go to very low-temperatures is to study and use quantum-mechanical effects.

I should also point out that not all materials would become superconducing even at 0 K, this can be shown theoretically.

 

[Amorphus]

yeah, but most materials become conductors at some degree.... rubber becomes a good conductor at, I believe, -230F (someone check me up on this)

remember, we cannot get anything to absolute zero. we've been able to get close, but thats the progress made by large labs with multimillion dollar budgets. the average end user will be saving up his entire life to pay for the bathrooms in that kind of facility. and it would be the mens' bathroom, too. after it had been used for a few years (you know what those are like... *shiver*)

also, remember that there is an innate speed at which processors run (which is why your overclocks sometimes are restricted by your processor, as opposed to the usual : RAM, northbridge, etc). why? because they simply cannot conduct the electricity rapidly enough to ensure stable operation - the peaks and the troughs of the electic signal are in the wrong range to properly function - some peaks don't quite reach the point at which the processor says "alright, theres a 1", and some troughs may slip above that point, causing the 0 to be identified as a 1.

what does that mean? that means that there is a definite point at which your processor will NOT go any further, and, on top of that, I would think that no matter how you do it, with current technology, you wouldn't be able to run a processor at any speed.

tough, eh?

 

[stardust]

you cpu wouldn' function if it was frozen at absolute zero. the gates wouldn't speed up. if the cpu were to function, your gates would be useless. also at absolute zero, nothing moves, that means to move ur gates, the temp would've increased dun u think?

 

[Howard]

Originally posted by: stardust
you cpu wouldn' function if it was frozen at absolute zero. the gates wouldn't speed up. if the cpu were to function, your gates would be useless. also at absolute zero, nothing moves, that means to move ur gates, the temp would've increased dun u think?

We don't need to hear it more than 5 times, and the type of language you exhibit is the first I've seen in HT.

 

[SuperTool]

The CPU would probably fail long before that.
The cycle time of the CPU is limited by the critical path. That's your maxtime constraint. That means you cannot run the CPU faster than your CP allows. That part will improve with lower temperature, but it won't be zero.
Additionally, there are mintime constraints. That means that path from one flip-flop to another must not be so short that the data races through the next stage flip flop before it closes. The CPU circuit designers design the chip to work within a certain window of constraints. So maybe we design for voltages between 0.7 and 2 Volts, and temperatures between say -30C and 100C, and various variations in transistors. All races are fixed for those constraints. But outside those constraints, we don't test the circuits, and we don't design for those cases. It might work, or it might not work. But if you have 100 Million transistors, chances are good that something will fail. So it's important to make sure that you pick the right chip for the application. You probably don't want to take a P4 to space where the temperatures range from cold to hot and where it's going to be bombarded by radiation.