Why doesn't water and liquid nitrogen cooling make a huge difference in oc

science man · Jun 27, 2009

Hey guys I don't understand why ok for example they're (intel, AMD) to make a quad to go to 3.2GHz with a fan right? Now just think about about for a second. Think of how physically powerful a fan really is. Now compare that to water cooling. Nice freezing water. Think of how much of an enporvment that is physically. If you truely understand that then don't you question why water cooling can't get the cpu up to 50GHz? If not, why? And knowing all that why can't liquid nitrogen get that cpu up to 1THz? These are my true questions for you overclockers.

Mods I'm sorry if I put this thread in the wrong sub-forum. If you feel the needed to move it be my guest.

Exempt · Jun 27, 2009

I think your location is wrong.

dark bishop · Jun 27, 2009

um, the chips have deficiencies in them that limit them, its like sometimes you get a chip that will go from 2ghz to 4ghz on air, other times you get on that will only go to 3ghz. water can only remove so much heat, no matter how much you have, yes having a larger resivour will lower temps by 1 or 2 degrees but that's still a limited amount, i could put a computer in a closed loop that has a million gallons and moves it all in an hour but only so much heat can be transferred still. liquid nitrogen is unfeasible because it would take a ridiculous amount to run it for any extended amount of time, it gets far colder than water because thats a physical property of it, water doesnt have anywhere near that ability.

we have the cooling solutions to keep chips stable at certain temperatures for extended periods of time, but the chips themselves dont have the ability.

science man · Jun 27, 2009

dark bishop said:
um, the chips have deficiencies in them that limit them, its like sometimes you get a chip that will go from 2ghz to 4ghz on air, other times you get on that will only go to 3ghz. water can only remove so much heat, no matter how much you have, yes having a larger resivour will lower temps by 1 or 2 degrees but that's still a limited amount, i could put a computer in a closed loop that has a million gallons and moves it all in an hour but only so much heat can be transferred still. liquid nitrogen is unfeasible because it would take a ridiculous amount to run it for any extended amount of time, it gets far colder than water because thats a physical property of it, water doesnt have anywhere near that ability.

we have the cooling solutions to keep chips stable at certain temperatures for extended periods of time, but the chips themselves dont have the ability.

But the water and liquid nitrogen were regulated like a carborator and if their travel was from the cpu to a freezer that wsouyld refreez them to their orginal state of matter?

CryptokiD · Jun 27, 2009

science man said:
But the water and liquid nitrogen were regulated like a carborator and if their travel was from the cpu to a freezer that wsouyld refreez them to their orginal state of matter?

what you basically described is how an air conditioner or refrigerator works.

in order for water (or anything really) to significantly move heat it has to phase change. since water boils at different pressures (and since you dont want a 100c cpu) you would need to run watercooling and a vaccuum to lower the temperature at which it boils to something more reasonable like 30C, 20C etc.

liquid nitrogen is good for temporary cooling because it has a super low boiling point. so it sits there boiling (phase change) until it's gone. im no liquid nitrogen expert but i suppose you could recapture the gas, compress and turn it back into a liquid. thats exactly what an a/c system does, only with refrigerant and not liquid nitrogen, but the basics are there.

dark bishop · Jun 27, 2009

your talking of a phase change system that uses a refrigerant to cool something by changing from liquid to gas, then having a compressor change it back to liquid. these sytems work but they cost quite a bit and take a fair amount of power. depending on how many stages it is it can still only get so low.

science man · Jun 27, 2009

So if you did water cooling with a vaccum could u get an infinte amount of GHz?

Bobnova · Jun 27, 2009

Not at all, the silicon can only change it's conductive state so fast, and even the electrons have a peak speed that can't be exceeded.

jediman · Jun 27, 2009

like everyone is saying. . . temperature is only one part of the operation of semiconductor devices

ChanceCoats123 · Jun 27, 2009

science man said:
So if you did water cooling with a vaccum could u get an infinte amount of GHz?

The processor itself is the problem. You could use a very large amount of water and continuously add water so that the amount of heat energy entering the body of water is basically the same as the amount of space for that heat energy, but the chips themselves cannot infact go that fast. If you google the 500ghz transistor, you will get a better idea. IBM took a single transistor and made it actuate at 500 billion times a second. A processor obviously has more than one transistor which would take a very complicated control system in order to actuate all some odd million transistors in turn and still be able to do things correctly with any sort of stability. That is where new processes and technologies come in. Silicon cannot physically do that. Once todays processes get down to around 11nanometers, 11 nanometers between transistors, there will be a point where the electrons will simply jump from gate to gate instead of going through each gate. Which is every inefficient. And thus cannot work.

Aynjell · Jun 28, 2009

There are people who really push the limits of the available CPU's out there. One of the things that they run into is something I recall them calling "arcing", where the switches are actually being pushed so hard that the electricity is just going right over them (causing them to not work right or at all). Basically, it's like adding a turbo to a car. If you set it for 30PSI on a stock engine, I give you 5 city blocks before you throw a rod if that.

If you put it to the test, things tend to fail. Following my previous example, water cooling is like pulling out your stock rotation assembly and going with forged equipment instead of cast or the compressed parts some engines are made with.

science man · Jun 28, 2009

ok I was about to google more materials for a cpu other than silicon and then I thought, what about fiber optic glass? (I'm still gonna google but still)
EDIT: I found this but still my question remains abvout fiber optic glass being a canidate.

DOUBLE EDIT: hey I googled optic computing and it said that it was basically the use of light instead of ectric, but wouldn't you have to have some kind of solid at least to produce the light? I don't get it.

ChanceCoats123 · Jun 28, 2009

I really have no idea on the fiber optic computing, but I would like to mention this:

By 2007, the industry may have to shift to a new kind of insulating material called high-k dielectrics to reduce overheating power levels as chips get smaller and the circuitry is jammed tighter and tighter. High-k has been demonstrated but isn't yet proven in mainstream manufacturing.

This actually made me laugh, at 8 in the morning, because Amd still uses SOI and Amd chips run cooler without a doubt... :screwy:

But just to add on to what anyjell said, the water cooling a replacement for one on the weak links in the system as a new forged crank will replace one weak spot in an engine, but then you have other things to deal with. When you start building more power, you also need more fuel and thus the fuel delivery system is brought into play; a higher gpm pump is required, larger jets in the carb or better flowing injectors. I guess you just need to understand it is not one element of the processor or cooling, it is a combination of so many things going on together.

SCFAdamVcG · Jun 28, 2009

because of the flow speed
depending on how fast it is the cooler the cpu will go,probably the higher the oc if it can be chilled quick enough
the pump works a little like a fan in as much as rpm
i hope this answers your question..
also liquid nitrogen has a reacton where as water doesn't.

N3buchan3zzar · Jun 28, 2009

Your answer lies in the physical design and characteristics of the processor itself.

Both Intel and AMD realized that keeping up with Moore's Law was becoming too cost prohibitive, hence the move to multi-core processors.

I remember when people were predicting 10Ghz CPUs but that isn't likely to happen anytime soon. The records of 8+Ghz are done with older Pentiums and yet they do not perform as well as newer architectures at lower clock speeds.

Most processors can only function at low temperatures up to a certain point, referred to as Cold Bug or Cold Boot. An exception at this time is the Phenom II which seems to have no known cold bug point. Guys like K1ngp1n have even used liquid helium to get it colder than liquid nitrogen can take it.

As a cpu is colder, it is more efficient in it's use of electricity. That is why people can achieve higher clock speeds with lower voltages using such methods as phase change cooling or LN2. It's a double edged sword. To get higher clock speeds requires more electricity which generates more heat which lowers the efficiency of the cpu and requires more electricity which generates more heat...ad infinitum.

1.5v on air will generally net you less Mhz top speed than 1.5v on water which is less than 1.5v on phase which is less than 1.5 on LN2.

The engine analogy was pretty good. There is simply a physical limit to how much stress your cpu(engine) can take in Mhz(rpms) and it will fail or break going past that point.

The gates simply have a limit to how many times they can be switched on and off....

science man · Jun 29, 2009

SCFAdamVcG said:
because of the flow speed
depending on how fast it is the cooler the cpu will go,probably the higher the oc if it can be chilled quick enough
the pump works a little like a fan in as much as rpm
i hope this answers your question..
also liquid nitrogen has a reacton where as water doesn't.

N3buchan3zzar said:
Your answer lies in the physical design and characteristics of the processor itself.

Both Intel and AMD realized that keeping up with Moore's Law was becoming too cost prohibitive, hence the move to multi-core processors.

I remember when people were predicting 10Ghz CPUs but that isn't likely to happen anytime soon. The records of 8+Ghz are done with older Pentiums and yet they do not perform as well as newer architectures at lower clock speeds.

Most processors can only function at low temperatures up to a certain point, referred to as Cold Bug or Cold Boot. An exception at this time is the Phenom II which seems to have no known cold bug point. Guys like K1ngp1n have even used liquid helium to get it colder than liquid nitrogen can take it.

As a cpu is colder, it is more efficient in it's use of electricity. That is why people can achieve higher clock speeds with lower voltages using such methods as phase change cooling or LN2. It's a double edged sword. To get higher clock speeds requires more electricity which generates more heat which lowers the efficiency of the cpu and requires more electricity which generates more heat...ad infinitum.

1.5v on air will generally net you less Mhz top speed than 1.5v on water which is less than 1.5v on phase which is less than 1.5 on LN2.

The engine analogy was pretty good. There is simply a physical limit to how much stress your cpu(engine) can take in Mhz(rpms) and it will fail or break going past that point.

The gates simply have a limit to how many times they can be switched on and off....

except for the end of N3bucher's post most of this gives backup to my hypothsies which is that u can get an infinite amount of ghz if u kept adding liquid helium and volts or at least to an extreme since it sounds like u'd eventually stress out the silicon of the cpu itself.

NoL · Jun 29, 2009

How do they at all? Quite the contrary.

Tsnowflake · Jun 29, 2009

The engine analogy was pretty good. There is simply a physical limit to how much stress your cpu(engine) can take in Mhz(rpms) and it will fail or break going past that point.

The gates simply have a limit to how many times they can be switched on and off....

this is your answer, temperature is only one of the limiting factors.

axis01 · Jun 29, 2009

That's not even considering that a MB has it's limits as well.

Axis

science man · Jun 29, 2009

Tsnowflake said:
this is your answer, temperature is only one of the limiting factors.

yep that would be the extreme which ibm so far has proven it to be 500ghz.

Why doesn't water and liquid nitrogen cooling make a huge difference in oc

Disabled

Member

Member

Disabled

Member

Member

Disabled

Senior Member

Member

d20 in a jacket

Member

Disabled

d20 in a jacket

Member

Member

Disabled

Member

Member

Member

Disabled

Similar threads