Spinning Heatsink Cools CPUs 30 Times More Efficiently

[CAT-THE-FIFTH]

Supposedly:

http://gizmodo.com/5920224/brilliant...re-efficiently

[Sysadm]

CPU goes pop if it fails? Sounds like a stupid idea.

[CivilDrone]

pretty mad idea. water cooling already exists

[DanceswithUnix]

CPU goes pop if it fails? Sounds like a stupid idea.

Why would this design cause it to go pop any more than fan failure on a conventional HS+fan?

I presume the thin air layer continues to be as good/bad as the TIM layer they are replacing, or maybe the HS is flying in which case it might park in contact.

Also if the main bearing is air, it is simple to replace before it wears out making it potentially more reliable

Anyone know if heatpipes still work when spinning at 2000 rpm? Might limit the capability if it is just plain metal.

[Sysadm]

Why would this design cause it to go pop any more than fan failure on a conventional HS+fan?

I presume the thin air layer continues to be as good/bad as the TIM layer they are replacing, or maybe the HS is flying in which case it might park in contact.

Also if the main bearing is air, it is simple to replace before it wears out making it potentially more reliable

Anyone know if heatpipes still work when spinning at 2000 rpm? Might limit the capability if it is just plain metal.

I can't see the heat dissipating from the CPU as fast as direct contact of TIM and a heatsink. Try connecting a CPU to a motherboard without a heatsink - It just goes pop in about 2 seconds regardless of any thermal shutdown. Perhaps it does work but I won't be trying it until I know for sure it works, especially on overclocked systems!

[Willzzz]

Well I'm happy to listen to the guys who understand fluid dynamics, generally held to be above rocket science and brain surgery in terms of complexity.

If it stops spinning it's still going to function as a heatsink, and there is no reason this would be any more likely to fail than a regular fan.

http://www.geek.com/articles/chips/s...ling-20120625/

[Pob255]

Try connecting a CPU to a motherboard without a heatsink - It just goes pop in about 2 seconds regardless of any thermal shutdown.

no it doesn't, unless you've actively turned off thermal shutdown in the BIOS, what happens is the cpu cycles will get cut as it starts to rapidly overheat.
leave it for 5min to cool down and you'll be able to turn it back on again, granted it'll rapidly overheat again.
Stick a cooler on it and it'll run happily.

If you manually disable overheat shutdown, then yes the cpu will keep getting hotter until the silicon shatters. But why would you do that unless you are testing how hot it can get before the cpu goes pop?

I do have a few arguments to the article and it's claims.

It spins at just 2,000 RPM and sits a thousandth of an inch above the processor. Sandia claim this setup is extremely efficient at drawing heat away from the chip, in the order of 30x more efficient than your typical heatsink-fan setup. This is because it removes thermal resistance by cutting the amount of motionless air within the cooling setup due to all parts moving. As it spins quite slowly, you get the added benefit of noise reduction, too.

Since when the heck is 2000rpm "quite slow"

hydrodynamic air bearing

hydrodynamics = The branch of science concerned with forces acting on or exerted by fluids
ok it may be referring to air acting like a fluid in limited volumes, but I still think that term is a bit suspect.

And as the whole unit spins, you aren’t going to get dust build up (ever).

This is blatant bull****, all you have to do is look at the blades of a fan, you'll find they get covered in a layer of dust. You'll not get the fluff matting that you tend to get with stock coolers which is a major factor in messing up coolers, but that's not the same as saying you'll never get dust build up.

EDIT: they talk much about improving the design of the of the fins for maximum air flow vs surface area and cooling, but they really don't say much about how the heat is transferred into it.
it seems to be a large flat base plate that impeller heatsink sits on.
These two faces are going to be have to be ground very smooth to keep the very small air cushion between them.
Granted that's a large contact area between to two halves but I connect help but question how much resistance is being added by the air cushion.
If you really care about noise and dust that much then passive would be a better way to go.

[Willzzz]

The article itself may not have fully understood the idea, but the attached video provides a much better explanation.

The noise reduction isn't due to the rpm, but the fact that it is a single unit, not forcing moving air through a static heatsink.

I suspect the dust thing is similar, in a conventional cooler dust builds up on the static parts and is then transferred onto the fan.

[Agent]

Try connecting a CPU to a motherboard without a heatsink - It just goes pop in about 2 seconds regardless of any thermal shutdown. Perhaps it does work but I won't be trying it until I know for sure it works, especially on overclocked systems!

Absolute nonsense. How can you make an absolute statement about it going pop, only to say "Perhaps it does work" in the next sentence?

All modern CPUs throttle and then shutdown before they get damaged. AMD has had this in since at least the XP/MP days (The older Athlon used to fail spectacularly), and Intel since at least the P3 days.

You don't need to test it, just Google / Youtube it.

http://www.youtube.com/watch?v=gkrMjX8RwYk (XP / MP surviving)
http://www.youtube.com/watch?v=06MYYB9bl70 (P3 + 4 surviving and old AMD Athlon (pre XP) dying)

[Agent]

no it doesn't, unless you've actively turned off thermal shutdown in the BIOS, what happens is the cpu cycles will get cut as it starts to rapidly overheat....

If you manually disable overheat shutdown, then yes the cpu will keep getting hotter until the silicon shatters..

Unless this is changed in the latest generation, both Intel and AMD CPUs will just stop doing calculations and hard lock the machine and have done for a very long time.

The thermal shutdown you're talking about in the BIOS is at the motherboard level, not at the CPU level which also has additional protection.

edit - some info from the old mobile duo range on thermal shutoff: http://www.intel.com/technology/itj/...sign_point.htm

[scaryjim]

... (The older Athlon used to fail spectacularly) ...

I once figured I'd "check" if an old Athlon was working by firing it up quickly without a heatsink, getting to BIOS, then shutting down again. I figured I should have half a minute or so to play with. It popped within ~3 seconds, with a rather smokey whiff...

[Pob255]

The article itself may not have fully understood the idea, but the attached video provides a much better explanation.

The noise reduction isn't due to the rpm, but the fact that it is a single unit, not forcing moving air through a static heatsink.

I suspect the dust thing is similar, in a conventional cooler dust builds up on the static parts and is then transferred onto the fan.

I did watch it and I did say I have issue with the article
In the Video they say it's "resistant to dust fouling" and "doesn't collect much dust" which is not the same as saying "you aren’t going to get dust build up (ever)"
Saying fan dust buildup is somehow blow back from a cooler heatsink is wrong as you'll get dust buildup on all the fans in a pc case, even intake fans without anything behind them.
It would be more appropriate to say "the design minimises dust build up to such a degree that it should never become a problem."
The idea that these would make a better cooler than an basic aluminium heatsink with a fan attached, I can agree with, but we already have them in the form of heatpipes and aluminium or copper sheet fins, which have a far greater surface area to mass ratio than an extruded aluminium heatsink.
Taking it a step further would be an area where I could see this being good, in coolers that currectly use impellers combined with heatpipes and sheet fins such as laptops and graphics cards.
Making the impeller fan itself an additional part of the heatsink would increase the overall cooler volume to surface area ratio, increasing the capacity of a cooler without adding to the size.

This not be the first time we've see someone come up with a better cooling concept, that's better than a basic stock cooler but not better or more economically viable than other existing cooling systems, and to me that's going to be key.

EDIT: oh and a fan or impeller makes noise reguardless of of pushing air through a heatsink or not, yes a heatsink can add a lot of turbulence to the air greatly increasing the noise, again stock heatsinks are not great on this issue, but your basic heatpipe tower cooler is far far better.

[Agent]

I once figured I'd "check" if an old Athlon was working by firing it up quickly without a heatsink, getting to BIOS, then shutting down again. I figured I should have half a minute or so to play with. It popped within ~3 seconds, with a rather smokey whiff...

It was basically instant on those chips. They didn't even have a heatspreader which guaranteed crispiness in a few seconds

[Time Bandit]

There is a link to the paper here from Jan 2010 for those who are interested - its from a US DOE research lab.

I can quite be fagged to go through all the detailed assumptions and calculations but the language is not that of an independent, peer reviewed publication; from p14, fig 9:

"THE PROBLEM WITH CONVENTIONAL HEAT EXCHANGER TECHNOLOGY: All opportunities to improve heat exchanger performance have been exhausted"

hmm -bold statement, along the lines of iron ships will never float, man will never fly etc. I am happy to be swayed by independent evidence, but in the meantime would add a couple of things:

• It looks like TIM is still used to mate the processor to the flat plate on which the impeller spins, so no saving there

• Air is not frictionless - stick your finger on the end of a bicycle pump and pump it hard (I think the heat sink could be dissipating rather a lot of its own heat!)

I may be missing some of the dynamic effects here, but the thermal conductivity of air (~0.025Wm^-1K^-1) is about 8200 times less than that of aluminium (~205Wm^-1K^-1) and about 15400times less than Copper (~385Wm^-1K^-1) yet the article says;

"The convective mixing provided by this shearing effect provides a several-fold increase in thermal conductivity of the air in the gap"

The dynamic effects must be considerable.

[kalniel]

There is a link to the paper here from Jan 2010 for those who are interested - its from a US DOE research lab.

Correction: it's from the company that are marketing this fan design. They are a spin off/funded by DoE.

Jeffrey P. Koplow
Remote Sensing and Energetic Materials
Sandia National Laboratories