Slashdot Mirror


Doomsday PC-Cooling With Dual-Cascade Coolers

An anonymous reader submits "Four (4) compressors cooling one PC! Yes, it's big, yes it's heavy, yes it's loud and yes it does get your CPU and GPU cold, very cold. Is -100C cold enough for you? Cascade cooling is yet another chapter in a Finnish overclocker's neverending quest for optimal PC performance. Those things go down to -80 to -100C and can maintain the temperature. See here for the whole article with the pictures of the project."

17 of 311 comments (clear)

  1. Time = Money by Anonymous Coward · · Score: 1, Informative

    If you wait and actually put the PC to some practical use, instead of modifying, you can get a faster PC for the price you paid for the original PC plus Time and Modifications.

  2. Re:minimum temperature by frdmfghtr · · Score: 5, Informative

    when you have an accident with liquid nitrogen you lose a finger. when you have an accident with liquid helium, you lose parts of the neighborhood.

    You're thinking of liquid hydrogen, not helium. Liquid helium is damn cold, yes, but it won't explode.

    --
    Government's idea of a balanced budget: take money from the right pocket to balance...oh who am I kidding?
  3. Re:In Finland... by apharov · · Score: 3, Informative

    Sorry to disappoint you with this, but it really doesn't work that way... While the temperatures would be quite about the same as in this refrigerator the terrible weather would kill any computer in a few seconds ;)

    Seriously though, often in winter when I have had to melt my freezer / refrigerator I've just put the foodstuffs in a box on the windowsill and opened the ventilation window. Everything stays frozen for at least long enough to to get the extra ice out of the freezer. I'm saying this as a Finn of course :)

  4. Forget cold feet... by WIAKywbfatw · · Score: 3, Informative

    Summary: Four (4) compressors cooling one PC! Yes its big, yes its heavy, yes its loud and YES IT DOES GET YOUR CPU AND GPU COLD - VERY COLD - EVEN DAMN COLD! Is -100C cold enough for you?

    Forget the cold feet, it's going deaf from the noise all that cooling generates that is your real problem. What's the point of having a PC that's so loud that you need to wear ear mufflers to be able to use it or else risk losing your hearing?

    Being able to hear yourself think while you work or hear the in-game audio while you play is a good thing.

    --

    "Accept that some days you are the pigeon, and some days you are the statue." - David Brent, Wernham Hogg
  5. Re:Not Quite by viniosity · · Score: 5, Informative

    Impossible due to the first law of thermodynamics which states that energy can be neither created nor destroyed. Here's a good link to brush up on your thermo101:

    http://www.geog.ouc.bc.ca/physgeog/contents/6e.h tm l

  6. 4.4 ghz by cmacmanus · · Score: 2, Informative

    So, he was able to obtain somewhere around 4.4 GHZ on a 3GHZ Pentium setup, with XP to boot [blegh]. Not bad, but give me a year and I'll have the same thing at not such an inane temp. and size. Admirable project, nonetheless.

  7. Article Text by Anonymous Coward · · Score: 5, Informative


    Summary: Four (4) compressors cooling one PC! Yes its big, yes its heavy, yes its loud and YES IT DOES GET YOUR CPU AND GPU COLD - VERY COLD - EVEN DAMN COLD! Is -100C cold enough for you?

    Intro

    If you are an overclocker you know that keeping things cool is the key for big clockspeeds. The cooler that comes with the CPU ain't going to get you very far. It must be replaced with better cooling if you want to get really high clocks out of your hardware. But what is good enough? Even the biggest and baddest heatsinks won't get your temps much colder - in other words they wont give you much extra in terms of MHz. Watercooling is a nice option cos it has huge cooling capacity but does it really give you a big gain in CPU speed? Usually no because it can't get colder than the air cooling the water. So what can you do if you really need to get more speed out of your system??

    Vaporphase cooling is the answer here. Vaporphase cooling is what keeps your freezer and ice cream cold. Vaporphase cooling is what 'all the xtreme-overclockers' are using nowadays. Several people have noticed that going from +40C to -40C makes quite a difference in CPU overclocking potential (talking about 200-600MHz here). There are even commercial solutions that go all the way down to -40C and even a bit colder. If you feel that you must get one of these just go to nVentiv website, check who is your local reseller and get one

    But what if you are a real speedfreak and -40C ain't cold enough for you?? Well there is always dryice (-79C) or Liquid Nitrogen (-196C) or even liquid Helium (-268.6 C) for you but the problem is that its not really possible to get constant CPU cooling with these. LN2 and helium are actually too cold for your little CPU - it just wont operate properly at such low temps.

    But you know those low-temp freezers they use in labs? Those that go down to like -80..-100C and can maintain the temperature. Good temps for CPU cooling eh?
    These are cascade vaporphase coolers. They are called cascades because of multiple cooling stages (normally two). First stage uses 'normal' refrigerant like R404 or R507 and cools down to around -40C. The second stage uses a special low temp refrigerant like R23 or SUVA95 or R1150 and can get the temperature down to -100C level. The first stage evaporator is cooling down the condenser of the 2nd stage - this makes it possible to use a refrigerant with very low boiling point in the 2nd stage. Normal cascade design uses two compressors - one per stage. This also means that it is not a very compact cooler.

    Here is a picture of such a freezer (the door has been ripped off):

    Cascade cooling is yet another chapter in my neverending quest for optimal PC performance. I've tried quite a few cooling solutions already (waterchillers, peltiers, R404 vaporphase, dryice, ln2 etc.) but cascade vaporphase was something new to me.
    This time I was lucky enough to locate not only one but TWO cryofreezers - both were supposed to be broken - so I got them for free.
    The first one (the one in the picture on page 1) had problems with the system fan and because of that the owners decided to send it to the junkyard. True, there really was a problem with the fan. It didn't blow any air at all - but then again no power was coming to the fan powerconnector. I made external power input for the fan and it started to work nicely. With the freezer door closed it would get the inside temp down to -91C.
    Obviously it would do nothing for PC cooling in its original form so I had to convert it to a CPU cooler.

    Testing cascade stage 1 - its charged with R404 refrigerant and it went down to -40C.

    2nd stage parts installed - CPU cooler is ready for a test run.

    Its working! First test run got it down to almost -100C with no heatload. Pretty good with R23 refrigerant (boiling point @1bar = -82C).

    Here is a picture of the evaporator installed on P4 motherboard. I was using a 3GHz P4 CPU here and it would clock to around 3.6-3.7GHz with good heatsink. W

  8. Re:Why? by Idarubicin · · Score: 4, Informative
    And so it goes with the lower you go. At -100C you could probably keep your 3GHz PC up at around the 14GHz area, which is way faster than anything on the market. Isn't it worth the money on cooling to experience a slice of tomorrow?

    This works to a point...but there are physical limits to how fast a present-day chip can be overclocked. Aside from potential limits on how fast gates can switch inside the chips, you start running into concerns about things like the speed of light.

    At a hypothetical 15 GHz, light will travel 2 centimetres (about 0.8 inches) per clock cycle. The Pentium 4 die is about 1.7 cm on a side, with a 2.1 cm diagonal. If one corner of the chip needs to talk to the other corner of the chip, it will take a minimum of two clock cycles just for the signal to travel there and back--and that neglects the fact that electrons move slower than light. You can bet that Intel's designers didn't anticipate those kinds of timing issues.

    So you might be able to clock a 3 GHz chip to 4 or 4.5 GHz with this setup, and do so much more stably with less wear and tear on the chip--but you're not going to be able to scale the chip's speed to 10 or 15 GHz.

    --
    ~Idarubicin
  9. Mirror by evn · · Score: 5, Informative
    The site is a little slow so I've mirrored the low resolution (~640x480) images and text right here

    Go easy on her, it's on my ISPs web space. Wait a minute, I'm still upset about not have truely unlimited access so on second thought: bag on it! ;)

  10. Re:Someday by T9D · · Score: 2, Informative

    > Liquid nitrogen is diamagnetic, meaning it is slightly affected by magnetic fields. In other words, the coolant could possibly conduct enough electricity to cause damage the hardware. This is not true. That it is diamagnetic means it slightly repels magnetic fields. This has no impact on its electrical conductivity.

  11. Re:Someday by ImTwoSlick · · Score: 2, Informative
    don't think you could do that. Liquid nitrogen is diamagnetic, meaning it is slightly affected by magnetic fields. In other words, the coolant could possibly conduct enough electricity to cause damage the hardware.

    (Affected by magnetism) != conductivity

  12. Re:Why? by barawn · · Score: 3, Informative

    You can bet that Intel's designers didn't anticipate those kinds of timing issues.

    Actually, just to nitpick: they did.

    There are two (maybe three? I can't remember off the top of my head) pipeline stages in the P4 which are "drive" stages - stages where the pipeline does nothing except wait for data to travel from one side of the chip to another.

    It was part of Intel's design to allow the clockspeed of the chips to reach ~ 10 GHz.

    Now, that being said, that's the fundamental design, not the current design. Almost definitely a modern chip can't run even twice as fast as its rated clock speed. The chance of the margins being that large are just zero.

  13. Re:The finnish ... by Anonymous Coward · · Score: 1, Informative
    What's the finnish word for ... compensating?


    "kompensaatioliiketoimi", or a word starting with "korva", depending on what you're compensating.
  14. Re:Not Quite --- wrong. by Anonymous Coward · · Score: 1, Informative
    Q: what does the fact that the total amount of energy in the universe is a constant have to do with the theoretical possibility of offsetting global warming?

    A: nothing. Parent is not Informative.

  15. Re:In Finland... by You're+All+Wrong · · Score: 2, Informative

    Don't do this if they're sandblasting your building.
    One PSU & one CD-R drive are now sitting somewhere in a Finnish landfill because of grit.

    YAW.

    --
    Your head of state is a corrupt weasel, I hope you're happy.
  16. More Wronger by Anonymous Coward · · Score: 2, Informative

    Although liquid helium is the coldest known substance, it has a very low, ah, what, 'heat of vaporisation'. Meaning it takes a very small amount of heat to raise a given amount of liquid helium to its boiling point. So that, even though it's very very cold, it makes a rather inefficient refrigerant.

    I'm going back 35+ years for this, back to the days I was a kid interested in cryogenics and spilled a whole thermos bottle of dry ice and alcohol slush on my hand. Screw that, I took up model rocketry (launched parallel to the ground.)

  17. Re:The finnish ... by Fizzl · · Score: 2, Informative

    Don't know what meaning of the word are you searching for. So here's few:

    kompensoida = compensate
    hyvittaa = compensate as in reimburse
    korvata = almost, but now quite, a synonym for hyvittaa.
    kumota = compensate as in neutralize

    have fun :)