question on subzero temps and cpus

OblivionLord

Limp Gawd
Joined
Apr 16, 2004
Messages
168
If the threshold temp of a cpu is 60c and anything beyond depending on how far you go causes damage.. then whats the technical reason for cpu's to misscalculate or whatnot when they are under subzero temps?
 
Well.....at subzero tempetures, things start to slow down. I recall a research team at GA tech has been trying to simulate absolute zero, and they say some crazy things happen at very cold degrees.

I would think that at a cold enough temperature, the transistors might lag or even not work, thus creating errors.
 
Why would some of the people on water have a limited CPU frequency at much lower load temps then an idle temp of good air cooling?

I know that every cpu is different but since in this scenerio we arent looking at subzero temps.. why would there still be a limit?
 
Where are you getting the proof for this? I would just like to see this, cause I wouldn't think you would have frequency problems unless you get to a really cold temp. I wouldn't think any normal water cooling setup could get the CPU to a temp that would stop it from working correctly or limit the frequency at all.

The watercooling usually doesn't lower the idle temp so much as it lowers the load temp. This means that it should have nothing to do with limiting frequency, it should just keep the CPU from getting warmer when it is working.
 
Approaching absolute zero won't be a problem for most of us so in general your processor will be faster at low temperature. There can be situations where there are timing violations at real low temperature. It may clock faster but there are errors. Not likely but possible.
 
what im saying is that on air the best you can do with the TT extreme 120 if thats all your using is in the high 40s whereas on water a really goo system will get you in the low 40s to high 30s..

What im asking is why on water or even air for that manner is there a limitation for the processor to only reach soo high of a clock if the temp on isn't reaching the Thermal specification of the chip which on Q6600's its 95c on the G0? Of course there are other bottleneck factos but, if everything else in on the mobo is cool like the chipsets, ram, case then whats to prevent the cpu from reaching higher speeds?
 
Most of the time I would say it is the chip itself. Even with dry ice and such, it is not easy to get much higher. You have to have a really good chip. And keep in mind, those water cooling and air setups are not capped. If you give the CPU enough juice(voltage) it will continue to knock the temperature up.

I am pretty sure that is actually how some of those guys with extreme cooling do it. I am sure the guy who got the Q6600 to 5gigs cranked up the voltage way higher than a WC or air unit could handle.
 
what would the issue be internally with the chip that heat isnt an issue where it just won't obtain higher clock speeds.
 
Well think about it, can you run further/faster inside an air conditioned gym, or outside in 100 degree heat.

Just like you, transistors are built to withstand a certain amount of heat(which comes from the power/work that is put into it). The heat that is being dissapated as energy. If you blow dry ice onto the transistors, you can run more power/work into them.

Water and air cooling does not make the transistors colder, it just dissapates the heat so that it isn't running as hot. It would be like standing in front of a fan or having water splashed on you. Your body is still at the same core temperature, it just isn't going up.

Keep in mind though, putting a CPU under dry ice and ramping the voltage up is going to drastically kill the lifespan of the chip, and like I said before, not every chip is going to be able to handle the voltage even with dry ice.

And to answer your question, the quality and setup of the silicon I believe is the reason some chips are unable to handle the clock speeds. Someone correct me if I am wrong though.
 
what would the issue be internally with the chip that heat isnt an issue where it just won't obtain higher clock speeds.

It has nothing to do with temperature. It has to do with voltage. It can only run so fast at a given driving voltage. Up the voltage too high and you risk damaging the processor, even with extreme cooling solutions like phase change or LN2. That's how extreme overclockers kill chips. After a certain point small flaws in the chip start to cause issues and you won't be able to run higher clocks with any form of cooling at all, so as others have said, you have to get lucky and get a really good chip.
 
Since all the switch elements are some form of FET the speed is directly related to capacitance on the outputs and the transistor's channel resistance. Increasing the supply voltage reduces the channel resistance improving the switching speed. This is why higher voltage gives a stronger signal. That's not to say having the higher voltage has no other benefits to speed. Interestingly, lower voltages mean the switch doesn't need to be as fast if everything else is equal.
 
Back
Top