Slashdot Mirror
Ask Slashdot: What Would Happen If You Were To Put a Computer Inside a Fridge?
dryriver writes: This is not asking what would happen if you were to place your iMac inside your kitchen fridge. Rather, what if a computer casing for a high-powered graphics workstation with multiple CPUs and GPUs, lets say, worked just like a small fridge or freezer, cooling your hardware down without using any CPU fans or liquid cooling and similar. How much would such a fridge-casing cost to make and buy, how much electricity would it consume, how much bigger would it be than a normal PC casing, and would it be a practical solution to the problem of keeping high-powered computer hardware cool for extended periods of time? Bonus question: Is such a thing as a fridge-casing or "Fridgeputer" sold anywhere on the world market right now? Linus Tech Tips tackled this question in a video a couple of years ago, titled "PC Build in a Fridge - Does it Work?"
try a meat locker!
I'm sure you can find the formulas online. Not sure how they scale though.
Refrigerators for food can ice-up and otherwise have problems with condensation. Any refrigerator sufficiently advanced to have features to avoid this will cost far more than the equipment needed to deal with waste heat specifically for computer applications.
Additionally, most inexpensive consumer-grade refrigerators are not really oriented toward dealing with constant heat. Most food cools and remains cool once it's in-place, and when hot food is put into a consumer-grade refrigerator it takes some time to really come down to the internal ambient temp. Expensive consumer-grade refrigerators may be equipped to better cool hot food quickly, but they probably are not geared toward continuous heat.
If the original purpose of this was to get a dorm or cubicle fridge and build a computer into it, I would not recommend doing that. The difference in air temperature between the inside of the fridge and the ambient is not great enough relative to the waste heat produced by the computer to justify the build.
Do not look into laser with remaining eye.
When you cool a volume of air you squeeze all the water out of it. The water condenses on the computer components, and your computer breaks.
I suppose as long as you dehydrated the inside first and kept several moisture absorption packs in there it would be ok.
Honestly water cooling is going to be better bang for the buck. you're just going to do an inefficient job of what the ambient air does fine to normal functional components.
bend like the reed
https://what-if.xkcd.com/155/
"worked just like a small fridge or freezer"
"without using any CPU fans or liquid cooling and similar."
You're simply moving the liquid cooling slightly further away. Liquid cooling IS involved if you have a fridge involved or "similar".
"Give orange me give eat orange me eat orange give me eat orange give me you." -Nim Chimpsky
https://www.youtube.com/watch?v=B8bhGw4vUFE
*rimshot*
Thank you folks, I'll be here all night.
The thermal flux capability of a home refrigerator is tiny. And, in fact, refrigerators are extremely well insulated. Your 1000W mining rig would easily overpower the thermal capabilities of the fridge, and you'd have a smoking crater.
Except this is liquid cooling (~the refrigerant), it's just coupled very ineffectively to the CPU via air instead of direct contact. This refrigerator also makes use of a phase transition, unlike conventional water cooling (unless you're getting really toasty!), but it's not unheard of to build this into your rig.
But again, just blowing cold air is going to be rather inefficient for a single computer (data center is another issue I guess). Not to mention, you will need a *larger* radiator than you would otherwise, as refrigerators of course "make more heat than they make cold," and they just pipe the heat off to the air. So...the TDP of your case has now gone up by adding this unit.
A fridge is an efficient machine for a smaller amount of work. A fridge works via a compressor motor that isn't on all the time which actually generates more external heat than it internally cools when used inside a room. Compressor motors as such are not designed for continuous operation but even if they were compressors are not designed to remove that much generated heat continuously. such a compressor would be huge and likely many times the size of the space it was trying to cool and would require an order of magnitude more power to operate and hence the room it operated in would be like an oven.
The reason a fridge is efficient for cooling food is that that box is insulated and so the compressor does not need to be on most of the time.
Not into the whole INOF thing, you make a dumb device that connects to the internet I will either disable your internet access or, hell, disable your internet access.
There was an XKCD what-if about toasters vs freezers. While a CPU doesn't run as hot as a toaster, the end result should be pretty much the same: inside the fridge ends up hotter than outside. https://what-if.xkcd.com/155/ Conduction through air is poor at removing waste heat. And you don't get much convection in an enclosed space.
Democracy is two wolves and a sheep voting on lunch.
You're going to need air circulation, whether you rely on the fridge fan or your cpu fan. You might be able to rely on convection currents to pull air through your cpu heatsink. Maybe you could get away with a giant heat sink (1ft square?) to service a single cpu. A refrigerator typically cools things and keeps them cold, as opposed to cooling things that are constantly cooking (as in a PC). A typical freezer is likely the same. You're giving up cheap cooling (fan on heatsink, ambient air) for a more expensive method (pumping all of the heat out of the refrigerator).
The surrounding air being a few degrees cooler will make very little difference since the heat in a computer is not spread out but in a few very small dense areas.
Cooling liquid works because it is like a fridge, but the cooling happens right up against the hot parts. Cooling the whole case from the outside-in would requite temperatures that would be very expensive to create, and probably require competently different materials for cables and boards as most plastics and metals do very poorly at -100C.
Troll is not a replacement for I disagree.
nt
Air is an insulator, so placing computer parts in an air-mass inside a cooling vehicle (direct gas-expansion refrigerator) creates a built-in inefficiency that makes no good sense fiscally or energy wise.
In contrast oil conducts heat beautifully and a computer immersed in mineral oil offloads heat amazingly well. https://www.pugetsystems.com/s...
Secondly, not all parts of a computer generate heat, so putting the whole case in the air-cooled fridge is just an exercise in inefficiency.
Far better to do what the overclockers do, which is to take the heat-generating parts, [in some case de-lid them and put different thermal-conductive paste on them] and put a coolant directly on the components. You don't need super-custom components that can sit in a mineral-oil bath, have the liquid-cooled setup deskside where you game (or run power apps like booting Windows) and the cost is nonprohibitive.
But hey, if you want to stick a PC in a fridge, go right ahead. People do stupid things all the time. Just take a look at the current POTUS.
E
Cooling stuff in a fridge filled by our atmosphere will condense water around the components especially on humid days. Components will last a short while before failing from corrosion or a short. Running a setup like this (without air gap leaks, must be completely sealed) for a week will work if the air inside is cycled very rapidly to slow or stop condensation forming. Without adequate airflow or other dehumidifying methods, this setup will fail. Another possible way to prevent condensation forming is to bath the containing area with high vibrational ultrasound, or to remove the oxygen and fill the fridge with a noble gas.
In a Slashdot first the summary answers its own fucking question!?!!!!
Your average computer can likely reject more heat to room-temperature ambient air than a refrigerator can move from the inside to the outside.
All from quick internet searches.....
Assuming a 1kW power supply at a 100% load, you need to move 1kW of heat outside the fridge just to maintain temperature.
Assuming a coefficient of performance of 4, you need 250 watts of electrical consumption just to hold the current temperature. The guy in the linked video used a dorm fridge, lets see if I can find some power draw specs on one of those....
No, I can't... but someone tested their dorm fridge and found a running power consumption of 145 watts. So if your dorm fridge is significantly more powerful than the random forum guy's random fridge, you might be able to maintain the fridge temperature equal to the ambient temperature (like the guy in the video did for a bit).
A walk-in restaurant freezer would likely have the cooling capacity to allow you to try the effect of colder air on your machine- or you could just stick it outside for a while on a cold winter day.
Alcohol, Tobacco and Firearms should be the name of a store, not a government agency.
For reasons stated above, putting a computer in the fridge isn't going to have the desired outcome. But, I have to wonder why this is something you would even ponder doing. Closed loop liquid cooling is cheap and very effective. If you want something more exotic than that, submerge it in Flourinert.
Wow, I haven't seen a true Karma Whore Troll in a long time...
Not sure a refrigerator is designed to have a heat source constantly working against its cooling system.
Problem 1, cooling the air lowers the amount of water it can hold causing condensation, requiring either a dehumidifer or keeping the container sealed to prevent rust.
Problem 2, most refrigerator type devices are not designed to be running non-stop, by constantly cooling an object that is generating heat you overwork the motor.
Problem 3, refrigerators are loud, louder than most pc cooling fans, and way less energy efficent.
Problem 4, cold air alone is not enough for sufficent CPU cooling, you would still need a heat sink and fan most likely to avoid excessive localized heat, unless conditions were artic, which leads to...
Problem 5, most common computer components are not designed to be operated in those conditions long peroids of time if at all depending on how cold you get.
Problem 6, Occam's razor, why? Coolness of something different and hard?
Problem 7, a working setup would probably cost more than a decent liquid cooling system and weigh more.
And on and on.
I don't remember the name though and I didn't think much of it either. This was when people were experimenting with Peltier cooling, and watercooling tended te be entirely custom jobs.
Alright I remembered the Vapochill name.
https://www.hardocp.com/articl...
Quite understandably maybe, it puts a fridge in your computer rather than your computer in a fridge.
You get a much larger amount of thermal fluctuation, which is worse than running it with crap quality fans. You'll get condensation all over the place which is worse than running it with crap quality fans. As Alton Brown has said, fridges are designed to keep cold things cold, not make hot things cold. Yes people have tried it, because every single person who builds their own systems has had this idea. Companies have tried and failed to meet this consumer demand with bloated $700 monstrosities that perform......worse than crap quality fans.
http://www.grcooling.com/ its what we do here http://vsc.ac.at/systems/vsc-3..., very out of date website sadly
Stupid idea... would make your beer warm.
1000W of thermal equates to roughly 3,400 BTU. Yes, you can get window AC units that are easily 10x this and keep your rig cold. Keep in mind that those run R22 (or equiv), with a lower working temp of the mid 30's (F).
You would like something that could handle 5,000 BTU with R12 (or equiv) - which can get about 60deg F colder. You need that colder ambient air to be able to cool the chips without fans. You will still need heat sinks. Air is a lousy coolant medium using only convection.
Enjoy the warm beer.
People do indeed use refrigeration for PC cooling, but not in this way. They're referred to as Peltier coolers, and have been used by hardcore overclockers for many years. The idea is that the cooler itself is a small solid-state device that, when a current is passed through it, creates a temperature difference between one side and the other. You place the cold side on the CPU and the warm side connects to the heat sink (or other cooling systems).
Peltiers have many difficulties, however. They add total heat to the system, so you really need good case ventilation (or a water-cooled system). Without that efficient ventilation, your system ends up being hotter rather than cooler. Plus, as many people mentioned above with the refrigerator, they can have condensation issues if the "cold" side of the Peltier winds up below room temperature.
Peltiers can help, but in the end they are limited in the same way that the refrigerator case is limited: air is a terrible conductor of heat.
Well-designed water-cooled systems can easily surpass Peltiers or the refrigeration idea posted here, as water is a far superior conductor of heat.
There are DYIs and commerical ones out there, Here is the first one I found. Computer Liquid Cooling (Submersion)
;)
If you do DYI just besure to get it right
What are you trying to achieve? A fridge could be useful, but why would you then specify no CPU fans or water cooling? No fans is good for silence, but this only holds if your fridge is also silent.
I would expect that if you've gone to all the trouble of making a refrigerated case, you'd be foolish not to use water cooling to make best use of your refrigeration. If you're determined to use air cooling, you'd be best to make it closed-cycle, otherwise you'd have to deal with condensation and ice build up due to humidity in the air intake.
A kitchen fridge is designed to run intermittently. If this was your cooling technology, you'd probably want to add some thermal mass to the system, else its on-off cycles would be very fast.
Quattuor res in hoc mundo sanctae sunt: libri, liberi, libertas et liberalitas.
Duh
Datacenters have air conditioners to keep their servers from frying. But hardware costs are falling while energy costs rise. So these days some datacenters are running at 90F because it's cheaper to replace anything that fries than pay higher power bill.
But your home fridge is not going to cut it because it's not rated for food that continuously generates heat. If you want, you can mount a desktop enclosure onto a window A/C unit.
That is the only way I could keep my Amiga 4000 running in the summer. The only problem is that I couldn't fully close the fridge.
Not going to work, at least for very long. Although fridges can get pretty cold, it takes time. They don't actually transfer heat quickly enough.
You're probably asking in the wrong place... There are dedicated places for crazy enthusiasts who do this sort of thing. It's been around for a while. Here is one such place
And yes, it can be purchased commercially.
Stick your computer in front of an air conditioner instead?
Damn Gavron, calm the fuck down! Guys just asking a question!
Suggest laying off donuts and coffee for breakfast.
In college I build a PC case around a window unit that was wired to run non stop. It worked quite well though when the power went out it would get immediate condensation. On the up side, it was great for keeping your drinks cold too. On the down side, I needed a jacket to gsme.
look. this isn't a widely used idea because it is stupid. most of the time you could use it only for the cpu if it was cheaper to overclock a cheap chip than to buy a more expensive chip. most people who come up with this idea don't actually know what a fridge does, so they abstract the fridge into a magic box. after you look what a fridge does, this seems less ideal. after you look at the benefits you get from going from ambient 23 celsius to 3 celsius, it seems even more stupid. after you look at what it takes to proof it against condensation and ice, it starts to look even more stupid.
but back to the dorm fridges. lets say your fridge has an efficiency of 50%. lets say your pc consumers 400 watts of power in gaming. your fridge has to be consuming 800 watts while you play and NOT MANY FRIDGES CAN DO THAT - your typical fridge cannot cope with a constant heat load that big - it will be running full time and STILL the temperature inside the insulated box will keep going up.
in addition your electricity bill will go up so much that you would be better off buying a better gpu anyways, even if you had an industrial freezer with enough btu to keep your rig cool.
in addition, you would have to dump the heat outside. so really you would be better off just buying an aircon unit in the first place and using that to cool water and use that to cool your pc components if you want to go extreme - or just pipe the aircon unit into the box - because THAT IS THE CHEAPEST WAY to buy enough capacity that can handle the heat load from the computer.
and really it's not worth it now, there was a time when you would get gains enough to be kind of worth it(dollarwise), but not nowadays.
world was created 5 seconds before this post as it is.
The important element is actually the thermal mass of the refrigerator, which allows the compressor to cycle. You could do the same thing for a computer, but it is an extra piece to make it work-- roughly 5 gallons of water in the fridge would be enough mass for a 200W computer.
it wouldn't affect at all. the fridge is an insulated box. your 5 gallons of water would only help if you were occasionally using the computer and having it in hibernate rest of the time. it doesn't help with the base thing that a normal fridge compressor cannot cope with the constant heat generated inside the fridge and it will run constantly and the heat inside the box will keep going up.
having more thermalmass only makes it happen slightly slower.
world was created 5 seconds before this post as it is.
Condensation would happen.
It's why putting a PC in a fridge is rarely done.
Sure and you could "just ask a question" about putting ether in your gas tank for better gas mileage.
It's stupid.
Just like putting a PC in a fridge. Abysmally stupid.
I was happy to point out the science of why that is, as well as alternatives if you truly want to get good cooling.
If I was a troll I'd have mentioned that getting the cords out the cooling chamber would compromise the cooling, but, hey, I don't think that's of any significance given how much this whole idea sucks.
E
Whores? Dude, whatever you're smoking, your brain's fried to mush. Also if you haven't seen shit in a long time... lay off the crack and go outside and take a look. Loser.
How about just an air tight box with a block of dry ice to keep him company?
Walk out the door. Look slightly left and you'll see Moe's old fridge. It's been there since he was "evicted" by the local mobsters. Shame he fell terminally ill during the process. Throw out anything decaying and take it. Don't take Mimi's fridge. That SOB never worked and was thrown out for that reason.
Now take the computer you stole and rip out anything identifying it. Also get rid of fingerprints. Hope the owner that dies in a pool of blood never noted the serial numbers of each and every component. Fully clean the drives. You know the drill, use dd.
Punch a hole in the side of the fridge for cabling.
Put the computer inside the fridge.
Fire up fridge.
Close door.
Boot computer.
Close door again.
Observe.
Post findings.
Have lasagne your granny made.
Feel good about life.
Think about Moe with fake sentiment.
I hadn't the slightest objection to his spending his time planning massacres for the bourgeoisie... (P.G. Wodehouse)
Something like the Calyos NSG - S0
http://www.calyos-tm.com/calyo...
Domestic spying is now "Benign Information Gathering"
Once upon a time (late 1990s), a company called Kryotech essentially put a refrigerator inside a computer rather than do it the other way around. At the bottom of a very tall and heavy case, they had the workings of a refrigerator (compressor, condenser, coils, etc.) that pumped coolant to the CPU. AnandTech and Tom's Hardware still have their coverage on their sites, complete with diagrams.
In my refrigerator, it would have to negotiate for shelf space with the mary jane yogurt left over from Jerry Garcia's memorial service. I just can't bear to throw that out, ya know? It might get slimed by the bag of shredded cabbage that's just oozing everywhere. That has a "best if eaten before" date from the Bush years. HAW that is. There's a package of bacon I've kept as a momento of my college years. One of these days I'll have to throw that out and do penance for wasting perfectly good bacon.
The Russians have won. They have made the world a cesspool of distrust, greed, fear and hate.
Not sure if they are still around , but in 2011 Asetec released their second refrigerateD case.
http://www.trustedreviews.com/...
Microsoft, Apple, Google, Amazon what's the difference? All steal money from devs and control with walled gardens.
What happens if you put a computer inside a fridge? It gets hacked, of course:
https://www.youtube.com/watch?...
Linus Tech Tips actually did a video on this exact subject a while back and it actually works very well besides the condensation issues with a very small TDP computer. They basically did a test with a small ITX platform and then again with a full size ATX build. I would post the Youtube link but it's blocked on work computers.
The ITX platform stayed very cold as there wasn't much heat the dissipate. With the full size ATX build however, it couldn't keep up and actually started to get hotter than when in open air conditions due to the fridge walls acting as an insulator.
Have you ever fallen asleep at the keybhanusdiog?
What Would Happen If You Were To Put a Computer Inside a Fridge?
Your mom would be cross. And tell you to get back in the basement.
Refrigerators are not designed to deal with things that have a few hundred Watts continuous heat output. They are designed to cool things that are not actively producing heat down to about 5-7 degrees Celsius and keep them there using as little energy as possible.
No it would not work unless you take into account condensation, which I doubt that you could do well.
Yes, it would run for a bit. But I wouldn't see it lasting a single year.
Now now, creimer is an annoying mosquito of a human, but wishing death on him is not right.
Seriously?
Years old click bait videos are now news?
Server grade equipment has been specified with max air temperatures of at least 35C for over a decade (that's 95F for you metrically challenged third worlders). Modern stuff is even higher rated.
But really in the home just buy a water cooled PC for crying out loud.
You would die instantly.
I see most reasons were already mentioned.
I don't if somebody already tried this, but you could use a computer immersed in paraffin oil and cool that with fridge tech, if you just want to lower the working temperature.
Or you could just buy a large CO2 container and spray that onto the hot parts, but don't inhale too deeply for too long. :-)
Think of the task, a small surface area (chip) that needs to shed a lot of heat by radiation.
How quickly and how much energy (heat/infrared radiation) can the small surface pass to the surrounding atmosphere (fridge interior)? There's actually a physical limit to the amount of radiation from a perfect black body and a chip is not a perfect black body.
So for a small low temp chip set, a computer in a fridge would be fine.
For a modern high performance GPU, not so much.
What a fan and liquid coolant do is move the energy from the chip to a much larger surface area complemented by convection (moving air between the plates).
"Consensus" in science is _always_ a political construct.
So cool.
Air is a poor conductor of thermal energy so you would still need a CPU heat sink / cooler fan. In that case, it would cool the CPU slightly more efficiently due to the increased delta in temperature. If you were trying to run the computer without the CPU heat sink / fan, your CPU would burn up slightly slower than it would at room temperature because the air, no matter how cold it is, just cant conduct the heat away from the CPU fast enough.
Stupid questions like this have ruined /.
Dude! Or Dudette!
Are you that hurting for content? click bait BS about Pakistan, this article that anyone with basic knowledge of physics would find ridiculous... I mean its almost as if you are trying to get a job at buzzfeed.
I get that slashdot is nothing more than a news aggregation site, but seriously i would rather see fewer posts than bs posts that are either filler or clickbait.. If post density is really concerning you then maybe do some real reporting?
Go back to high school Physics class and PAY ATTENTION this time! If you had, you'd know about things like heat transfer, condensation, and so forth...
This is DUMB on so many levels it isn't funny...
I think the big mistake that Linus Tech Tips made was that they put the computer power supply inside the fridge. When I did it, I cut a hole for the power supply and drive cables and mounted those on the outside so that the only thing inside the fridge were the motherboard and video card. Condensation was not a problem. I mounted my motherboard on the ceiling of the fridge so that any condensation could drip down, but none ever accumulated. The computer components are warmer than the rest of the interior of the fridge, so any condensation will happen on the condenser coil of the fridge and not the electronics. I did have to increase the capacity of the fridge though. I used a cheap "beverage center" with a glass door that wasn't really designed to get to very cold temperatures, so to increase it's capacity, I bought two of them and pulled all the condenser, evaporator, compressor equipment from one and added it to the other. My goal was to be able to overclock the computer and also keep ice-cold beverages in there simultaneously. It turned out that I could do one or the other, but not both. If I overclocked, then I couldn't consistently keep the temperature near freezing. It was a fun project and was arguably successful, but ultimately, there are better ways to overclock and better ways to chill your beverages.
The computer heater would win, and the refrigerator/freezer would just stay warm. In fact, it would overheat because the fridge would be insulated and trap all the heat inside.
https://what-if.xkcd.com/155/
Popisms.com - Connecting pop culture
Inside a fridge? No that will no work they are closed systems meant for cooling items that do not produce heat. However there are phase cooling systems out there that work on the same principle as the fridge. LD-Cooling provides a pretty good phase change case. However phase change provide extreme cooling and has the same issue other posters have pointed out, condensation. Where you have a large change in temperature you will have condenstation. Linus Tech Tips actually did a pretty good build guide on this as well.
The thing a lot of bean-counters forget when they push to have the datacenter "run hot" is that things fail. In my experience, the thing that fails the most often in the datacenter is the air conditioner.
If you keep the datacenter well below the maximum ambient operating temperature of your computers, it will take an hour or two for the datacenter temperature to exceed that mark when an air handler fails. That gives you time to engage backup units, bring in portable backups, set up fans to redistribute air to the end of the room that isn't cooling, get a repair crew in, and so forth.
Or, you can save money by operating the datacenter right on the margin of what the equipment will tolerate. Then, when the cooling fails, you immediately have a temperature excursion beyond design limits. "Who cares? We've got redundancies and a service contract," says the beancounter. They get really pissy when you point out that the service contract states it's null and void if the computer is exposed to temperatures outside the manufacturer's design specifications...
Even worse is when the failure happens in winter, and some bright spark decides the answer is to open the datacenter doors and set up fans. Yes, the temperature goes down. So does the relative humidity, like a stone, and now you've got failures from static discharge...
Sure, you could get more reliable chillers. But that's a capital expense, and it's a facilities project—and like any modern corporation you've outsourced your facilities maintenance, and possibly sold your building to lease it back. So you've got to get those folks engaged, plus any necessary permits, plus engineering studies, and contractors for the AC, and possibly plumbers and electricians, and then the beancounters will ask why you aren't buying the stuff from the low bidder... and isn't the air cold in here already?
Current cooling: Heat sinks on hot components dissipate waste heat into a fluid (air). Circulation of fluid assisted by fans.
PC in fridge: Heat sinks dissipate waste heat into fluid (the air inside the fridge), which then transfers to another fluid (the refridgerant), and then dissipates it through another heatsink (the radiator on the back of the fridge).
This will be less efficient than the original method.
They make Phase-change cooling kits that have a dedicated cooling block that mounts to the CPU to literally refrigerate the CPU. you have to pack all of the surrounding area of the CPU with moisture blocking foam to eliminate condensation.
these types of systems cost a couple grand and up but they do help you get the CPU down close to freezing.
as far as practicality, unless you are trying to severely overclock your CPU for ePeen score, or your ambient air temp is 120+F, they have no real world use above what a standard watercooling system will do you for roughly $100.
Better question: What Would Happen If You Were To Put a Fridge Inside a Computer?
Question answered! http://www.instructables.com/id/Convert-a-Silicon-Graphics-Server-into-a-Fridge/
See how easy life could be if we just asked better questions?
It'll work fine, if you can get all of the water out of the air inside. If it needs to be opened, the system needs to be brought up above dew point before opening. You'll probably want desiccant packs (e.g. damprid) as well.
There is no XUL, only WebExtensions...
The reason why this isn't done is the same reason why it's not done for the engine in your car. Basically, there's no benefit of making your PC run _colder_ than the ambient air temperature. That's what a phase-transition heat-exchanger is good at. If you simply need your chips to not get significantly hotter than ambient air temperature, then it is more efficient to skip the compression, and merely rely on liquid cooling.
That said, phase-transition is not entirely unheard of in the realm of computing. It's merely a very niche application. Some aspects of quantum computers depend on materials behaving differently at very cold temperatures. They will use things like liquid nitrogen to achieve low temperatures. LN is produced by specialty compressors, but in essence, they behave the same as the compressor of a refrigerator. Conceivably, you could hook those compressors directly to the component in a cycle; this just isn't done because it doesn't serve any particular benefit when a lab can just purchase LN, as it's fairly cheap stuff.
Anyone remember this from back in the day?
http://totl.net/Eunuch/
Linus did this already: https://www.youtube.com/watch?...
If you have only one datacenter, you should be using cloud services. Networks fail a lot two and your latency would be really bad for customers on other continents. Even a second A/C unit is a fixed expense that deprecates slowly compared to constant utility charges.
what is with the all of the /. hate towards Creimer? or is this related to APK?
A mini fridge is still a hermetically sealable box that has slideable shelves, with an air gap filled with insulation between its inner compartment and its outer shell. Its size and shape makes it easy to adapt a housing for HEPA style filters to, and very large low rpm fans for high air flow and minimal noise. Modified for the installation of computer components, the result could be a very quiet, dust free casing for a computer.
Phase Change cases have been made for over a decade with overclocking in mind. VapoChill was a leader at the beginning of the last decade with OCZ CryoZ following suit spawning an entire subset of system modders who used phase change cooling. This spawned off the popular OverClocking (OC) website frozencpu.com. These were all sub >$500 dollars for a complete ATX solution.
A decade later, CPU speeds were no longer bottlenecking game performance and the advent of GPUs, and subsequently SLI functionality, which improved gaming frame rates and performance greatly, the need for crazy Phase Change systems diminished.
I mean, does anyone remember when OC Team Italy they used LN2 to overclock a Pentium 4 to 8.18ghz literally a decade ago?
To answer the question, yes, it's possible, yes, there are available solutions now, look up http://www.ldcooling.com/ they're in the >976,00 € range for the PC Block only. Is it practical? No, is it awesome? Yes.
There are other issues to watch out for. The preparation for the board includes sealing of the internal parts with silicone caulking to prevent condensation short circuits because only certain items like the GPU and CPU get put inside the loop and the surrounding areas will be cold enough for water to condenhttps://hardware.slashdot.org/story/17/06/12/2158239/ask-slashdot-what-would-happen-if-you-were-to-put-a-computer-inside-a-fridge?utm_source=rss1.0moreanon&utm_medium=feed#se and eventually freeze. Have to use foam ceiling around the CPU socket front and back as well Also you'll have to keep in mind that each off/on cycle will physically stress the components at the solder joints and even at the microscopic interconnect level.
Others have effectively covered some of the downsides:
- condensation;
- relative abilities to move heat.
Here are some more that I've not seen in this thread:
- the computer has point sources of heat generation. Typically, it is only worthwhile applying point cooling to the CPU and the GPU. Case-level cooling solutions can handle all the rest easily;
- that means that cooling the entire computer is wasteful, and I only mean that from a space management perspective;
- Want refrigerated cooling? Air or liquid cooling not 'cool' enough for you? They already exist! It's called an IceCap and it's a solid-state cooling system for CPUs. The design principle is the Peltier Effect, Google it. Why are you reinventing the wheel, except it's a not-very-popular wheel that not even 1/10,000 people uses?
- Still not 'cool' enough for you? Put your PC in a Data Center and benefit from the air conditioning there. Or, you know, simply run an A/C in your room...
- Still not 'cool' enough for you? Run dedicated refrigerated cooling pipes to the processor(s), like the old Cray 1/2 did. Expensive as hell.
The question might be fun from a naïve, beginner sort of perspective. However there are decades of research and engineering in this problem, and no one, NO ONE winds up putting a computer in a refrigerator. You kind of need to ask yourself why.
While not precisely the same, the concept of using a refrigerant was.
"The high-performance ECL circuitry generated considerable heat, and Cray's designers spent as much effort on the design of the refrigeration system as they did on the rest of the mechanical design. In this case, each circuit board was paired with a second, placed back to back with a sheet of copper between them. The copper sheet conducted heat to the edges of the cage, where liquid Freon running in stainless steel pipes drew it away to the cooling unit below the machine."
[This is a true semi-relevant story that I posted to Reddit a few years back.]
Many years ago, I got a field report of a few systems that were turning themselves off during the night without warning or seemingly a reason. We had some temperature-monitoring software that would automatically shut the system down if it got too hot, and the logs showed that was what was happening. But it was November and the systems were in snowy regions, so too-hot didn't make sense.
A quick look through the code showed me the problem. The programmer was reading the current temperature from the hardware as an unsigned byte instead of a signed one (as the data sheet for the part specified), so that an actual temperature of -1 C (which is crazy cold for a system) was being read as 255 C, and so it would shut down immediately. Later investigation found that the affected systems were in unheated loading dock areas.
The fix was, of course, easy, but testing it required putting the system in our kitchen freezer (with the various cables coming out for power, the monitor, and so on). The CEO walked in while I was sitting there with a keyboard, mouse, and monitor seemingly hooked up to the refrigerator and just said, "I don't think that I should ask what you're doing."
is this a troll question? tell me this is a troll question.