Theoretical Question on overclocks.

[Osmium]

     So I have been thinking on this for awhile. Since an increase in temperature on metal increases it's electrical resistance. From my understanding this is why chips at lower temps perform marginally better at than the same chip at a higher temp. Could keeping a chip at lets say 0 degrees celsius have a higher OC potential than a chip at 80 degrees celsius both at the same voltage? 

[beebskadoo]

This is the principal of LN2 cooling.

[SirRoderick]

It's incredibly insignificant. The reason LN2 cooling exists has nothing (or very little) to do with running temps as much as keeping the damn chip alive during INSANE overclocks. You're many times better off clocking the chip higher and getting "normal" temps than you are letting them sit chilled.

[Bluejay0]

One word. LN2

[Excon]

That is why exotic cooling exists.

[Osmium]

This is the principal of LN2 cooling.

 

I thought it was simply to keep it cool while you ramped up the voltage. I am just talking with temperature variance alone.

 

And I am not actually going to do anything with this information I am just seeing if I am correct. I am aware if you have super low temps you are supposed to OC them back to normal ish temps.

[Faa]

     So I have been thinking on this for awhile. Since an increase in temperature on metal increases it's electrical resistance. From my understanding this is why chips at lower temps perform marginally better at than the same chip at a higher temp. Could keeping a chip at lets say 0 degrees celsius have a higher OC potential than a chip at 80 degrees celsius both at the same voltage? 

It doesnt perform any better at lower temps, usually at high temps the cpu starts to throttle. And no OC potential depends on the quality of the chip, not on the operating temps.

[VSG]

It depends on both the temperature and the chip quality. For CPUs of course the chip quality (how high an overclock at a certain voltage) matters more though. For GPUs it is different - I can get higher stable overclocks at 15C than 25C ambients at the same voltage. This difference is exaggerated the colder you go- on LN2, people don't use a whole lot more volts than on water, but the overclocking is much higher.

For example, at 1.52v on water with an ambient of 15C and a load temperature of 40C core and 44C VRMs I can go up to 1552 MHz core on my GPUs. But the card under LN2 can go to 1800+ core at about 1.6-1.65v.

[LDG]

Not just increasing electrical resistance but also increases the leakage currents in the transistors inside thus creating some errors..

[slasher0161]

It has an effect but you have to consider the temperatures in Kelvin not Celsius or Fahrenheit. It more is in regards to clean power delivery though, the cleaner the supply the better chances of a good clock but the biggest limiting factor will always be the silicone wafer until such time as we use a new medium like carbon, then that will.