At what speed will processor heat output outstrip air cooling solutions???

[0roo0roo]

seems the heatsinks just keep getting bigger, will there be a point where air cooling will no longer be sufficient?

 

[Shalmanese]

Well, if we assume perfect heatsinks, and put an upper bound on how hot a processor can run, then its a simple problem on how much air can be moved with an acceptable noise level. Assuming 125 degrees as an unpper operating bound, we get approx 100 degree temperature differential between ambient and processor. From: http://www.cgd.ucar.edu/csm/models/cpl/cpl4.0/doc9.html, we get 1200J/m^3/K or 120KJ/m^3 of air. So moving 120KW/m^3/s or 56W/CFM. Then its s simple matter of seeing how low we can get a CFM/dB level.

edit: fixed maths.

 

[Mday]

Originally posted by: 0roo0roo
seems the heatsinks just keep getting bigger, will there be a point where air cooling will no longer be sufficient?

it's not a matter of speed of the processor. it's a matter of how many watts the processor will output.

 

[Peter]

Yes. There are quite a few variables.

Heat dissipation of heat source (CPU)
Thermal resistance of heatsink on both ends (against heat source, and against air)
Maximum temperature of heat source
Ambient temperature
Achievable airflow

If one of these items changes to the worse, you need to improve on one of the others.

(And then, of course, neither CPU performance nor clock speed are directly related to heat dissipation. So the original question is sort of missing the point.)

 

[beatle]

Also consider the shrinking die of the CPU. There is less surface area than ever before on CPUs if you disregard the IHS. At some point we will reach the limit of transistor size and we'll move to multiple cpus on the same die and/or package. This is mostly conjecture. Take with as much salt as you like.

 

[Peter]

Die shrinks affect thermal conduction toward the heatsink. It's #2 in my list On the other hand, they (usually) improve on the thermal output.

 

[mooncancook]

Liquid nitrogen colling is the future!

j/k. I think by then we'll have biochemical transistors that are so small and run so cool that heatsink is not even needed. it's all about nano-technology

 

[borealiss]

beatle brings up a good point. i think heat density is something that will push thermal conducting material development more than just power output of a cpu since copper/aluminum can only sweep away a certain amount of heat per square cm of thermal material. so you could have a 20 watt part burning up because the process and die is so small the heat density goes through the roof and the hsf cannot sweep it away fast enough.

 

[SuperTool]

I remember coming across Intel technical presentation slides a couple days ago describing liquid cooling for processors that looked like it's coming relatively soon.
Maybe Intel folks can comment on that.

 

[Mday]

Originally posted by: SuperTool
I remember coming across Intel technical presentation slides a couple days ago describing liquid cooling for processors that looked like it's coming relatively soon.
Maybe Intel folks can comment on that.

i doubt they can they dont wanna get fired

but it's obvious that effective water cooling or heatpiping techniques will be coming to the desktop market sooner rather than later.

 

[mooncancook]

I thought liquid cooling is already available. the problem with liquid cooling and conventional air cooling is that they are too bulky to fit in a thin notebook.

 

[SuperTool]

Originally posted by: Mday

Originally posted by: SuperTool
I remember coming across Intel technical presentation slides a couple days ago describing liquid cooling for processors that looked like it's coming relatively soon.
Maybe Intel folks can comment on that.

i doubt they can they dont wanna get fired

but it's obvious that effective water cooling or heatpiping techniques will be coming to the desktop market sooner rather than later.

It wasn't a confidential document, it was publicly presented at IDF. I don't work there, so I wouldn't see it otherwise without an NDA.
Here you go.

 

[Georgeisdead]

From a purly technical point of view, air cooling will be possible for some time to come. This is because it is simply a matter of using the fundamentals of heat transfer. Conduction and convection are the 2 most impotant modes of heat transfer here. Number one, you will eventually reach a limit on how much heat you can conduct from the die to the heat sink. Silver has the highest thermal conductivity of most materials barring exotic alloys which are pretty impracticle. Using the arctic silver thermal paste is obviously used to increase heat transfer to the heat sink by allowing any small void spaces on the die to be exposed to a very conductive medium of heat transfer.

I could talk all day about conduction but the real trick is getting all that heat dissipated. All those fins on the heat sinks are what make air cooling possible. Anyone will tell you that increasing the available surface area for heat transfer yields thousands of percent increases in heat transfer. Copper is not as conductive as silver, but an all silver heat sink might start getting pricy. However, reevaluating the geometry of the heat sink could result in a more optimal use of fins for further increases in efficiency. Convective heat transfer coefficients are also functions of velocity. It is possible in theory to get extremely high convective heat transfer coefficients by increasing the velocity of the surrounding air (i.e. through better case fans). However, this results in practical limitations since nobody wants to live or work in a wind tunnel.

But in summary, from a purly technical standpoint, it is possible through the use of exotic composite materials and manuipulating convective heat transfer of the surrounding air to keep air cooling at a pace where it can cool post-modern CPU's. Oh and by the way, lets dispel the statement right now that "it only depends on the speed of the processor not the watts". This statement is completely false and the person who posted it on the forums has had no formal engineering training ever. If he has had engineering training, he needs to go to the school that trained him and demand a refund.

I hope this gets some more people thinking about the abstraxt theoretical possibilities of air cooling.

 

[Georgeisdead]

I'd like to catch my mistake. The person quoted correctly that it IS the watts and NOT the speed of the processor. My mistake I read his statement wrong. He's fine.