Slashdot Mirror

← Back to Stories (view on slashdot.org)

Tiny Bubbles Key to Cooling Crazy Hot CPUs

Posted by Hemos on from the wick-the-heat-away dept.

Smaz writes "With future CPUs expected to generate as much as four times the heat of today's processors, wicking away that heat remains one of the biggest engineering hurdles in the biz. Researchers at Purdue have developed a pumpless liquid-cooling system that removes nearly six times more heat than existing systems. The trick, it seems, is in the tiny bubbles. From the Science Blog."

23 of 238 comments (clear)

  1. Pumpless circulation by stanmann · · Score: 4, Interesting

    I thought that with a properly pressurized closed system that convection and boiling would keep things cool enough. I know this isn't the first silent system, I'm just curious what special benefit the "tiny bubbles" and microchannels provide... unless we are going to another proprietary IBM standard bus.

    --
    Food not Bombs is a nice platitude but it breaks down when you notice that the Bombees are usually well fed
  2. Cavitation? by Ungrounded+Lightning · · Score: 5, Interesting

    It will be interesting to see if the shock waves from the cavitation (the sudden formation of the tiny bubbles) affects the operation of the chip or erodes the surface, limiting the life.

    --
    Bantam Dominique roosters crow a four-note song. Once you've heard it as "Happy BIRTHday" you can't NOT hear it that way
    1. Re:Cavitation? by iggymanz · · Score: 5, Interesting

      The sound waves from hard disks and power supply fans surely already make more vibration on the CPU than this would

    2. Re:Cavitation? by Guignol · · Score: 5, Interesting

      What your parent talks about is cavitation, the vibration you talk about is also a problem, but it has nothing to do with cavitation.
      What your parent reffered to was the formation of very tiny bubbles that quickly collapse and release microjets which are very damaging to surrounding surfaces.
      Those tiny bubles also have the (generaly) unwanted property of always orienting themselves so as to send the microjet against the surface of contact, thus making the problem more severe and less unlikely to happen that it might sound in a first thought.
      Those nasty microjets can do a lot of damage and are the reason why stainless steel helices of boats still get corroded.
      In the case of the proposed cooling system, the surface of the channels might be attacked by the released microjets until perforation, since it is so thin.

    3. Re:Cavitation? by Anonymous Coward · · Score: 1, Interesting

      This is an interesting point but I doubt that the bubbles would cause any erosion. Take inkjet printer technology, for example. Ink droplets are formed by using a heater to boil the ink and create a small bubble, which ejects a small volume of ink out the nozzle. This process can occur at 100KHz per nozzle.

    4. Re:Cavitation? by neitzsche · · Score: 2, Interesting

      The surface area they affect is also very tiny. Basically, you have a lot of tiny bubble doing serious dammage to a lot of very very small surfaces (that make one large surface.)

      Here's my horrible analogy: the starting surface is like a indi race trace - very smooth. After cavitation, the road looks like the Brooklyn Queens Expressway (where the surface area of the potholes is greater than the surface area of flat roadway.)

      --
      "God is dead." - Frederik Nietzsche
  3. Chips are expected to get hotter? by Anonymous Coward · · Score: 1, Interesting

    I guess that comes from somebody focusing on Intel and co, ignoring Transmeta...

  4. This brings me to my favorite rant... by mykepredko · · Score: 5, Interesting

    Where does the heat go?

    This seems like a nice technology to remove the heat from the CPU, but what I'm always wondering about is, where will the heat actually be dissipated into the environment? At some point, there has to be a heat exchanger where all this heat collected in the tiny bubbles is passed outside the unit. This is going to take a fair amount of space - one of these days we're going to see ads for heat exchangers that take up less space than the "standard" box available from Intel.

    I'm looking forward to a Beowolf cluster not only performing amazing calculations but also heating the building it's in.

    myke

    --
    Mimetics Inc. Twitter
  5. Next Generation Cooling by Strange+Ranger · · Score: 2, Interesting


    Now we know why Intel was so anxious to get their anti-overclocking technology working.

    --

    Operator, give me the number for 911!
  6. Fish tanks have used these for years by uiil · · Score: 2, Interesting

    There one of the cheapest filter methods out there. The bubbles drive the flow through an uptake tube of an already established siphon between the tank and the filter resivoir.
    The hardware layout would need to be orientation independant for a laptop though.

  7. something wrong by kEnder242 · · Score: 4, Interesting
    As liquid flows through the channels, it is heated by the chip and begins to boil, producing bubbles of vapor. Because the buoyant vapor bubbles are lighter than the liquid, they rise to the top of the tube, where they are cooled by a fan and condensed back into a liquid.
    I see two things that might be a problem

    -The chip needs to be at the boiling point of the liquid, maybe not a problem (freon anyone?).
    -What happens when the CPU isn't pointing up? (e.g. on a motherboard in a standard case) Will it overheat because the bubbles don't "rise"?
    --
    my associative arrays can kick your hash - TCL
  8. What happens when you tip the thing over? by flakk_jacket · · Score: 2, Interesting

    Since this relies on gas rising while a pool of dielectric fluid boils, I assume there's some air left in the system, right? So, what about when you have your laptop on an agle, and could this work with a traditional tower? It seems that for whatever application you use it in, the cooling unit would always need to be oriented "up".

    1. Re:What happens when you tip the thing over? by stanmann · · Score: 2, Interesting

      I would assume that it would work in a range of angles. And that you could attach one to a heatsink on a tower case, provided that there was clearance "above" the processor.

      --
      Food not Bombs is a nice platitude but it breaks down when you notice that the Bombees are usually well fed
  9. Re:yeah by glesga_kiss · · Score: 4, Interesting
    Bong makers are aware of this fact as well. Putting a bit of cloth/gause over the pipe at the bottom will make the air flow into lots of smaller bubbles, rather than a few large ones. Most surface area, more cooling.

    I'd love to submit an "Ask Slashdot" article on the making of bongs. I'm sure we'd see quite a few novel ideas from the MacGyver Smokers out there...

  10. So when can we stop wasting heat? by johny_qst · · Score: 2, Interesting

    Shouldn't there be some ideas to utilize a similar system coupled with a miniature sterling engine to get some of this energy back... regenerative braking is the only cool idea to come out of the automotive industry in the last couple decades of supposed innovation.

    --
    Fnord.sig
  11. Future CPUs will use LESS power by bhny · · Score: 3, Interesting

    I thought the whole emphasis on CPUs was changing from higher clock speed to lower power usage, even in servers. Google's number one requirement is low power usage in their servers.

    I'm sure the average PC in the future is going to be using LESS power than today.

  12. Suck the heat out? by Anonymous Coward · · Score: 1, Interesting

    How about lowering the pressure?

  13. efficiency question... by jaredcoleman · · Score: 2, Interesting
    ...developing a "pumpless" liquid-cooling system that removes nearly six times more heat than existing miniature pumpless liquid-cooling systems...

    Every comparison in the article was with current liquid systems. How much more efficient would this be than the heatsink/fan cooling my Athlon?
  14. Solid conductors by evilviper · · Score: 4, Interesting

    I don't see why there is so much effort on dispersing heat... It seems that the only reason systems have a fan is that it's the cheapest cooling method.

    Want silent cooling??? Design a case where the healt-sink goes from the processor, to the outer-shell of the case... Presto, no more restricted airflow, and no fans at all.

    Convection works well when there is a large surface area (unlike current CPU heatsinks), and there is little impediment to airflow (unlike current systems).

    In fact, you could have some incredibly hot systems if you designed a case with a large, EXTERNAL, healtsink, mounted so the top is flush with the case. It could look like a grill on the top of your case instead of a flat piece of metal, but be connected to the CPU with copper/aluminum.

    I've always been wondering why nobody designs computers that conduct the CPU heat outside the case. Anybody have some ideas?

    --
    Slashdot gets worse every day... Pipedot: News for nerds, without the corporate slant
  15. Heat pipes by Overzeetop · · Score: 2, Interesting

    That's what they are. Pretty standard effect. I'm guessing (from a scan of the article) that they've managed some magic concerning the microchannel interface, but the meat of the "discovery" seems to have been lost in favor of the amazing new heat-pipe phenomenon, which has only been around for thirty years.

    Here's an example:
    http://www.swales.com/products/heatpipes .html

    --
    Is it just my observation, or are there way too many stupid people in the world?
  16. New CPU architectures needed by Anonymous Coward · · Score: 2, Interesting

    The trouble with CPU manufacturers is that they are continuously increasing clockspeed to increase performance. All modern processors use CMOS (complementary metal oxide semiconductor) technology. CMOS is great in that the only time it uses substantial power is when the transistors are switched. Unfortunately, the higher the clockspeed, the more often transistors are switched, and the more power is consumed.

    New architectures are needed that can do a ton of work per clock cycle. Then, clockspeeds can be reduced greatly, along with power consumption.

    I heard an example one time that the human brain works at like 10Hz, and is capable of like 10^15 operations per second, but uses only 80W of power.

    I think late great physicist Richard Feynman drew up some equation to describe this relation, but I'm too lazy to look it up.

    Cheers!

  17. Cost of electricity.. by destiney · · Score: 3, Interesting


    The articles states:

    Innovative cooling systems will be needed in about three years for personal computers expected to contain microprocessor chips that will generate four times more heat than chips in current computers. Whereas current high-performance chips generate about 75 watts per square centimeter, chips in the near future will generate more than 300 watts per square centimeter, Mudawar said.

    Who can afford the electric bill to run such machines in their homes? I already stress over the few rooms in my house where I use 100 watt light bulbs instead of 60 or 75 watt bulbs. Can you imagine hooking up your shiney new PC in 2006, then getting an $800 electric bill the next month? Man..

    I guess powering down your system when not in use will become more common.

  18. lower power & different architecture processor by uwbbjai · · Score: 2, Interesting

    From what my university course taught me, CPU power is composed of 2 factors: dynamic power and static power, where dynamic power is dependent on clock speed and the other is independent of the clock speed. But dynamic power itself is the sum of the switching power (to charge up the transistors) and short circuit power (that split fraction of a second when both transistors are on, causing power to leak through). Both of these factors are directly proportional to the activity factor of the signal (the probability of a signal chaning from 1-to-0 or 0-to-1) The one signal that changes 100% of the time is the system clock. To distribute this one signal to all the individual components of the chip, a lot of power is wasted on generating the clock tree. Maybe we should seriously consider reviving the asynchronous CPU design. This would at least minimize the amount of signal activity. Besides, the faster the processor gets, the more time it spends in the NO_OP state, waiting for data to process. I say we should stop focusing on pumping higher clock rates and focus on other components that ARE TRULY THE BOTTLENECK. eg. memory and storage??? Or even use a different transistor technology, e.g. a CMOS transistor that recycles its charge to power other transistors?