Comparing Water Coolers: We Follow Your Lead

Lemur

Distinguished
Oct 5, 2007
2
0
18,510
The radiator is so small that good air-coolers beat it quite easily.

The sole benefit of water-cooling is that you can use a big radiator outside the case. This product does not really deliver anything useful.
 

T8RR8R

Distinguished
Feb 20, 2007
748
0
18,980
All water coolers and air coolers are limited to ambient temperatures. The biggest benefit behind water though is that it has a higher heat capacity than air, it also allows you to choose(to a point) the source of ambient temps. Where as air coolers always rely on the air that's in the case(unless ported and vented). What I mean is that there is nothing stopping you from dropping a radiator in a bucket of ice water or hanging it out your window in winter.

I do think that we're going to see more compact water cooling setups like the one by swiftech because there are alot of people that would like to have a water cooling system without having to do much more than read some instructions and snap it together. There is also no reason that these types of systems couldn't become more compact and more powerful overtime as well.

I think that high end air coolers perform just as well as low end water coolers, they are also cheaper, easier to setup(usually) and they don't require any maintenance other than a good dusting once in a while.
 

cleeve

Illustrious



Honestly, I thought the HyperTX 2 would have performed better, which is why I chose it for the review. I was surprised how hot it got. I'm pretty sure the temps were cooler in the Raidmax Smilodon case, which had alot more airflow but was a louder case than the Aurora. Unfortunately, the water cooloers required a 120mm fan space which the Smilodon couldn't accomodate... when the HyperTX tested poorly, I was out of time to do more testing with a higher end cooler as the article was already a bit late.

Having said that, the HyperTX isn't a bad cooler by any stretch of the imaginiation. I've seen reviews of the original Hyper TX vs. a Thermalright Ultra-120 - one of the best air coolers out there - and the temperature difference was only 5 or so degrees.

I've also seen a couple Ultra-120 reviews on the web, and from what I've seen the Ultra-120 gets around 67 degrees at 3.3 GHz. Yeah, it's a quad core but it's also almost 400 Mhz slower than what we tested here with the dual-core.

I'm not sure if Cooler Master did anything to increase performace of the HyperTX 2, but based on that I'd hypothesize that the Ultra-120 would have gotten in the 67 degree neighborhood under load in this case. But once again, with an air cooler alot depends on the case airflow...
 

03flat4

Distinguished
Jun 12, 2006
75
0
18,630
nice article. But you might as well get your hands dirty and get a H20-120 Premium which is the same price on Newegg. But, for ease of install I guess it's a good deal :)
 

joex444

Distinguished
Decent enough article. I think your choice to represent the air cooling department was particularly weak. You should have also used the air cooler in a typical air cooling environment -- if you would've used extra fans in the case, then put them in just to note that with air you could achieve such and such temperatures, but also do it with the same case configuration as the water cooled. Water cooling only really benefits from the additional air flow if it keeps ambient temperature down. And it should be lower by exactly that amount.

It's an interesting idea, just a waterblock with a small pump connected to a 120mm radiator. Very self contained. I don't think it qualifies as true water cooling, it's only cooling the CPU whereas more elaborate setups cool the northbridge and GPU as well. It also doesn't have any reservoir, which always lowers temperatures by making the water pass through the waterblock fewer times in a minute. Also, as pointed out, water cooling is supposed to have a large radiator outside the case, or atleast a large one using a few 5.25" bays. The advantage of outside the case is lower ambient temperatures, in addition to the obvious large radiator = more thermal dissipation. You can run 2 or 3 fans on the radiator instead of the 1 in this product.

Basically, this reviewed item is somewhat interesting, definetly a viable solution for more "off-the shelf" computers being so compact that the end user couldn't truly mess it up. For the price, air cooling makes much more sense. If instead of paying $10 for your HSF, you spent $60, you'd notice that the advantage this has over air cooling with a good heatsink is almost 0. For example, I'm running a C2D at 1.4V, but have it at 2.93GHz. If I'm not mistaken, watts is directly proportional to frequency, and squarely proportional to voltage. So, at the same voltage, the increase in frequency over 2.93GHz (that is the ratio 3.68/2.93) should increase the delta T over ambient. Now, my CPU idles at 45C in TAT, and loads at 60C. That's an increase of 15C, which if I ran at 3.68GHz @ 1.4V should be an increase of 18C over ambient during load. Thus, at load and 3.68GHz, with the same ambient temperature, I should expect to load at around 63C. Maybe a little higher. It should be atleast as good as the 66C reading you found for the item at 7v. Also, since your CPU was idling lower, at 42/43C, I can safely say that is due to a lower ambient temperature. Thus, if you had run the HSF I have (Tuniq Tower), I would expect the load temperatures at 3.68GHz to be 63-(45-3) = 60C. Exactly the same as the compact water cooler reviewed at 12v.

Yes, that's just a bunch of theory. If you disagree with my calculations or the theoretical background, please test a Tuniq Tower to see where it actually does fall.

If it does do 60C, then I have shown that for $150 you can get a tiny water cooler than does exactly what a $50 HSF does. You tell me what makes sense.
 

cleeve

Illustrious


Depends on your definition of "true" water cooling. To me, "true" water cooling would be a cooling system that uses water. I don't know of any industry standard that dictates what components must be present for a water cooling system. To me, it uses liquid as a coolant, hey... it's a water cooling system.





Well, BigBruin did a text that included the Tuniq Tower and the Hyper TX. The HyperTX performed better than the Tuniq Tower did on low speed, and only 5.5 degrees worse than the Tuniq on high:

http://www.bigbruin.com/2006/hypertx/large/chart2.gif


Based on this, I'd say the Tuniq would probably still deliver temps in the high 60's in this setup... and likely wouldn't have been able to match the 60 degrees of the Swiftech H20-120 @ 12v.
 

Elaruwan

Distinguished
Sep 20, 2007
2
0
18,510
I think you would find that a top end air-cooler (120 Ultra etc) would outperform this watercooling kit purely because the radiator is so small. If you compare the area of the fins in say the 120 ultra, and the simple 120mm rad seen here, the area on the air cooler will be larger, hence better cooling.

With the advent of heatpipes, the ability to move heat from a direct spot (ie CPU die) to a radiator has improved immensely, and so simple watercooling has lost its edge (Ie the 120mm versions that sit on the back wall of the case).

The ability to have really large radiators mounted somewhere inside the case that isn't directly above the CPU is waters strong point. Having a 2.120 or better yet a 3.120 sitting mounted to the roof of a case will see better results than straight air coolers. (Until they start sticking heatpipes out the side of the case with huge fin banks on the side.... DFI's Northbridge cooler anyone?)

Scary.
 

TRVertigo

Distinguished
Sep 14, 2002
12
0
18,510
I like the article as I do most.

I think it would be useful to modify the H2O. I may be wrong, but I bet the inside diameter of the outlet on the cooling block is the same as the inlet.

The outlet should be smaller by about 50%. This serves the same function as a thermostat in an automobile. The thermostat closes the flow of water out of the engine until the desired temperature is reached. Then it opens slightly to cool the water through the radiator.

The main benefit for the thermostat is it slows the flow of water allowing the molecules more time to come into contact with the load/heat and more time to lose the heat they have gained.

Reducing the outlet would serve the same as allowing the water to flow at its present rate and have an internal surface with fins in the cooling block. The fins/ridges increase the surface area and are expensive to cast. Then their function would also be improved by reducing the inside diameter of the outlet.

Water is a fluid and like air will heat under high pressure and cool under low pressure. Lowering the pressure on the outlet side may prove to give the H2O another 10% in efficiency.

Slowing the flow of water will also reduce the likely occurrence of cavitation in the pump by increasing the load.

I think it may be a fun hypothesis to test. The again the designers of the H20 may already have a smaller outlet. If so I was not informed.
 

ir_efrem

Distinguished
Dec 13, 2005
416
0
18,780
TRVertigo,

If you take out the thermostat from any auto you will NOT have heat inside the car. The purpose of the thermostat is to allow time for the engine and the water in the engine to warm up. No thermostat = a much much cooler running engine. Which is also a bad thing, simply because:

1. the water wont be warm enough in the heater core (small radiator that air blows over to warm up the cabin of the car)

2. Air and gas burn ALOT more efficiently at warmer temps. I can not give a quote but, the absolute best temp (for carborated engines anyhow) I believe is near 150 F. Fuel injected engines are more than likely a tad different due to the atomization of gas through the nozzles. Most thermostats I've seen have a temperature range (145 - 180 F)

The more water that passes through the engine the cooler it will be. I can not comment on how well this works in a water cooling system in a computer. But due to a few simple laws of thermodynamics I am apt to believe that the cooling depends on a couple simple things.

Amount of flow.
Surface area on both the radiator and cooling block.
Air flow over the radiator.

I think more can be done to the cooling blocks myself. Imagine a row of fins much like an air cooler inside the cooling block. This would increase the size of the cooling block but provide MUCH more contact area for heat to move from surface to water.

I don't mean to be argumentative about your comments but I think there is a small error in your fundamental logic where water flow is concerned.
 

TRVertigo

Distinguished
Sep 14, 2002
12
0
18,510
No argument taken.


You are so correct. The water has a chance to heat up because of the thermostat and engines do operate more effectively at about 150-195deg F. This is not only for the gas to burn but for the oil to work correctly.

Here we're not talking about the differences between the operation temps of engines and CPUs. We're talking about how heat energy moves between water and metal.

Anyway, the reason the water stays cooler with no thermostat is it races through the engine and forms a boundry layer thus insulating the majority of the water.

Let's say we're in a game of dodge ball. You represent water. The balls represent heat. You must cross from one side of the court to the opposite.

Will you get hit more running or walking?

May seem over simplified but, it's not.
 

Grub

Distinguished
Aug 30, 2002
2,814
0
20,780
Vertigo,

What your talking about is the differnce between turbulent flow, and laminar flow. I understand whetre you're going with this, but I think your just a little mixed up on the similarities of gases and liqid phases.

Liquids are, for all practical purposes, incompressible. Trying to expand them by restricting their flow (like the outlet 'nozzle' you propose on the cpu) wouldn't really work. Liquids just aren't going to give you the adiabatic heating and cooling (from compression and decompression) that gases provide.

However, you are correct in thinking that turbulent flow in the waterblock will increase heat transfer. This has been applied in the design of the old 'storm' waterblocks by swiftech. They called it 'pin impingement' IIRC. They had a small inlet nozzle that accelerated the liquid before it hit the waterblock's pins. SImilar to a high pressure washer. This creates lots of turbulent flow so that the mixing is better...better mixing = more efficient heat transfer.

However, you DO NOT want turbulent flow in your tubes. Turbulent flow increases the resistance to flow, thus placing a higher load on your pump. You want to save all your 'pump head', as its called, to pump that water through the waterblock and radiator.

Hope this iclears some stuff up for you. I've worked on heat exchanger R&D and high temperature ceramic reactors. If you have any questions I'd be happy to answer them.
 

ir_efrem

Distinguished
Dec 13, 2005
416
0
18,780
For the sake of what sounds logical I would agree, without seeing results. That turbulent water on the transfer surface would indeed increase cooling efficiency. The more water that comes in contact with the surface the better and turbulence would provide those means for sure. Many different means to cause even a small amount of turbulence come to mind. I still believe that a row of fins inside the block would be an excellent means for heat transfer regardless of the inlet/outlet size. They would also provide more turbulence as a side effect. With just a little R&D for this application would go a long ways I think.
 

grouchon

Distinguished
Oct 6, 2007
5
0
18,520


The primary advantage of this system over air cooling is that you can also add water-cooling to other components of your system. Cooling for your high end VGA card is certainly the most obvious candidate. Many people believe that this doubles the heat load, but in fact it doesn't. CPU and graphics are never fully loaded at the same time, except under artifical testing conditions. In real life, when you play a game, you VGA puts out heat in your system, while the CPU is around iddle, and when you run CPU intensive apps, then it's the opposite. Knowing that high end graphics run in the high 70's, and make quite a racket when the fan quicks in, and knowing that this system paired with a VGA water-block will keep your graphics card in the low 50"s with no added noise, you start realizing the true benefits of the unit.

Another advantage is the ability to upgrade with performance enhancing components if/when needed. The Radbox for example allows users for a mere $20 to install the radiator outside and to the back of the system, which may reduce temps by up to 10C because the radiator now uses ambient air instead of the heated air inside the case; it is also possible to add a second radiator to the system, which will lower the temps and or alow users to run practically silent by running the fans at 5 volts for example.

A heatsink has much more limited options. you can only change the fan if your temps do not satisfy you, and in most cases, this means higher CFM's and higher noise levels.

The argument consisting in questioning the benefits of liquid cooling by comparing this particular kit to high end heat pipes is fundamentally flawed. Replace the single 120mm radiator by a dual or a triple rad, and no air cooling in the world will ever compete. what I mean is that the real heart heart of the H20-compact kit is the Apogee Drive: a combination water-block + premium pump. This is the building block of the water-cooling system which will allow users to tailor their system to their needs. Finally, while the MSRP of this product is indeed $149, we think that street prices will also stabilize in the $139 range.

Gabriel Rouchon

Chairman, CTA

Swiftech
 

Houndsteeth

Distinguished
Jul 14, 2006
514
3
19,015
Mr. Rouchon,

Thank you for your reply. Just of note, while I appreciate your company and their efforts to market a better cooling solution, not everyone is on board with water cooling (I'm not one of them, as I have used your company's product, along with several other companies out there, for years now). Most of the overclocking world is stuck using air cooling because the overhead of getting into air cooling is very high, and the consequences of failing to do it right is even higher.

Those of us who have built water-cooled systems know this. A leak will spell the end of your system, and if you are a conscientious and honest consumer, you know that no electronics manufacturer would offer warranty repair for their product that was damaged by water damage from a leaking coolant system unless it was design to be liquid cooled in the first place. We void warranties on new mainboards and graphics cards everytime we remove the stock HSF and replace them with water blocks. It's not just the price of the cooling system that we put ourselves out for better temperatures, it's also any hope of product replacement should there actually be any manufacturing defect that is not detected before it is installed (at least, those of us who are honest, that is).

Hence, most people who watercool are the consumers who test hardware full bore before it is even modified for watercooling. Sure, we can buy prebuilt watercooled systems from almost any one of the big name integrators out there (Alienware or Falcon Norhtwest, for instance) and expect to get full warranty coverage, but you pay for it, and it's not nearly as fun as living through your own build experience and does nothing to further your own education.

I like that your company is looking at entry level products for the consumer. I fear that you will find that most people will shy away from your product because of the high cost of entry outside of the price-point of your product, and will steer towards cheaper, more conventional methods of cooling that don't require them to void their warranties. Leaks can and will happen, as every one of us watercoolers will attest.
 

TRVertigo

Distinguished
Sep 14, 2002
12
0
18,510



Still what I am talking about is take 2 identical skillets and heat them both to 400degF. Tilt both over a catch basin. Tilt one at 20deg and the second at 40deg. Pour equal amounts of equal temperature water across the skillets into the basins.

Which basin will have water with a higher temperature? Of course the one where the skillet was tilted to 20deg. Why because the water had more time to come into contact with the skillet. Hence my postulation that slowing the water across the block then allowing it to flow unimpeded through the radiator may improve the performance of a poorly designed water block.

My premise is that if the water is flowing too fast it will not pick up as much heat as slower moving water will. It doesn't have the time.

One of my degrees is in Civil Engineering. I'm not a fluid or cooling specialist at all. True water is basically incompressible, but it is uncompressible. That's how we have freeze drying. That is the reason for my remark about water as a fluid does cool when uncompressed.
 

grouchon

Distinguished
Oct 6, 2007
5
0
18,520


This is debatable. The DIY community is growing at an elevated rate due in most part to the availability of information. Granted, there is a limit to the DIY growth phenomenon, and that limit is defined by the interest people bear to building computers. But the amount of education, in other words the sophistication of DIY consummers is also accelerating, which goes in favor of putting together more complex/powerful systems. So I believe that there is a growing market for entry level good quality liquid cooling systems in the DIY comminuty. Just look at the number of new Vendors in this market, it speaks for itself.

The real question to me at least, is can a $50~$80 liquid cooling system be made that will clearly outcompete or even equal a high performance heat pipe. Based on the tests we have conducted in house the answer at this point in time is no, and this is why we chose to remain in the $140 price range: to preserve the performance advantage. On the other hand, with a huge increase in production created by OEM demand, costs are going to go down quite drastically. This will allow manufacturers to release more efficient systems at lower costs. So I do believe that we are getting there, one step at a time :)

Another advantage of liquid cooling that I forgot to mention earlier , particularly compared to high end heat pipes, is the ability to easily relocate the heat exchanger. This in itself is a major boon to OEM's who always have space constraints.

My simple conclusion is this: Liquid cooling is here to stay. Will it overcome the air cooling market, not anytime soon. Will it grab a share of that market as a sound thermal management solution, definitely.
 

Grub

Distinguished
Aug 30, 2002
2,814
0
20,780
To echo Mr Rouchon's enthusiasm for watercooling I'll add this...I got tired of buying a new top end air cooler every time a new socket came out. I had the best for socket A, then I had to buy the best for 939, then I had to spend again to get the best cooler for 775. Now I own the swiftech 220 kit. I can reuse all the components for as long as they last and the only thing I need to purchase when the next platform comes out is a waterblock. Granted, the waterblocks are usually the same as a top end air cooler, but I'm consistently getting better performance than the top end air cooler. SO the total cost of ownership over the long run is better with watercooling, IMO.
 

ir_efrem

Distinguished
Dec 13, 2005
416
0
18,780
TRVertigo,

I still believe there is a flaw in your thinking bud. In your latest example you are imagining an open ended system where the amount of available water is changing. The volume of water inside a water block never changes. Only flow can change. If you have some time take a look at the top performing water blocks, it's fairly nifty stuff. The inlet basically sprays water in much like a showerhead would spray water (submersed of course). Which adds a lot of turbulence.

I completely stand by the idea 100% that greatly increasing the flow would have very positive effects. Water hits the block, absorbs some heat and moves on, which allows cooler water to move in and do the same. If water stays there longer it heats up more and the cooler water waiting to flow in doesn't have as much of an opportunity to do it's work.

For a real world test... Take the same 2 skillets place one in bucket of water the other in a bucket of water with a hose stuck in it. Try it again with a hose in the first bucket with less flow. Think about the effects of more and less flow.
 

bobwya

Guest
May 21, 2005
692
0
18,980
Hi All,

Very interesting discussion on Thermodynamics... Actually more interesting than the original article but thats another story :)

I would just like to comment on a more mundane issue... Buying a full H2O kit is not NEARLY(!!) as good as selecting individual parts from different manufactures
(NB I am not thinking of homebrew parts here - lathe on :) )

I currently use a Thermochill PA120.3 in my dual Opty server and wouldn't dream of touching a Swiftech (or in fact any other comercial rad.) until they sort out their flow restriction and poor thermal transfer efficiency issues...

I do use Swiftech Apogee blocks as they provide good price performance as far as I can guage (they certainly work well in my system)...

The whole beauty of H2O is you fork out loads of dosh at the outset and then you only need to upgrade your GPU and CPU blocks as required. Plus maybe the odd pump replacement (since they only spec. 50000 hours normally)... Kits do not fulfill this as you end up with sh**ty rad that can't shift any heat from your water.
If you look at the heat transfer graphs on the Thermochill website you see that each rad. eventually maxes out at a certain Wattage that can be dissipated - however much you raise the flowrate or increase the fan CFM. A single fan PA120 can only dissapate about 200 Watts max before temperatures will start rising a whole lot above ambient in the loop. Thats fine if you are cooling a P-III perhaps :)

RE joex444 's comments
A reservoir generally serves the function of breaking up air bubbles. Unless it is specifically designed for heat transfer (TEC attached and/or metal with fins). It will not normally help 'remove' heat from a system. As all the water the loop and rad. will eventually heat up to +*C above ambient depending on the CPU/GPU loading.


Bob Wya
 

wtyrrel

Distinguished
Oct 26, 2007
4
0
18,510
Hello all:

While I have never experimented with water cooling, it had seemed appealing. I had previously looked at buying an Aquian ICM-509 from Koolance with a CPU heatsink you're ~$210. My main issue with water cooling is the price/performance. If you go with an outside cooler (the main complain of many posters) then the fans are outside, and therefore less muffled by the case, so end up being louder (A buddy of mine owns a Exos II system, also by Koolance). And the performance to me, is not justified. I'm running an older board with a Prescott P4 2.8; it's overclocked to 3.64 GHz still on stock voltages, but, as we all know, it's multiplier is locked, so my FSB is now running at 260 (1040). I have seen/read reviews of 'RAM water cooler kits' and most are quite unimpressive. Simply put, total air flow through the case is still important, is it not? With a well build case (Antec P180, in my case), and quiet 18db 120 fans rated at 75.8CFM from Thermaltake I have quite a lot of air flow through the case, which also keeps my northbridge cool and the flow of air directly over my HD's as they enter the case. I spent ~100 in air cooling, but all of it is transferable up, same, and in many ways, more so, than these water kits. (My air cpu cooler is a Thermaltake CL-P0024 with 2 of their 92mm fans (52 CFM per) blowing through it.) Oddly enough it runs 50C idle and 61C during a SisSandra CPU burn-in. Which is actually lower than it ran stressed with the stock cooler. The more I think about water cooling, the only real advantage is to truly OC, and I mean to the point where those weirdos start *making* voltage modifiers and applying them to their boards to squeeze that last drop out.

All this on an old 89W(?) P4, Sis shows it's wattage usage at an average 100W. Compared to the 65W Brisbane's or Core 2 Duos (which the CPU heatsink I have applies to both, as well) what could it pull off? I just think we've put too much thought into this, if we're going to spend $200 on water cooling, let's spend it on the air too and see where that gets us. The nice case with good airflow is going to be beneficial to either build (water or air) so I see no point in including that as part of my cost, and if we're going to spend $200 on cooling, then there's no point in skimping out on a weak case.

Any thoughts on what my cousin did? 5.5 CF mini-fridge. Put the whole case in there! Very quiet, runs at a (as he puts it) 'balmy 80F' under load and OC'ed. Mini-fridges run ~150-200.. better than either?
 

prcrlc

Distinguished
Oct 27, 2007
8
0
18,510
A more direct comparison price-wise would be Corsair's Nautilus 500 which is also a CPU-only system for $150.

I running a Nautilus with a D-Tek Fuzion waterblock. Kicks arse.

I agree that an H2O system can be either inside or outside the case, have one or more waterblocks, etc. The key word here is "water."

As a non-gamer (mine is a DAW), I only want increased stability so a CPU waterblock is necessary but Northbridge and GPU are not. Overclocking is counterproductive in a DAW and video processing is minimal.

BTW, I tried a Scyth Mine Rev B before I went with water. Today's coolers are so big and heavy that they flex the MB, and most importantly, often require repeated attempts to get them to seat well onto the CPU... Dinosaurs.

----------------------
MB: Gigabyte GA-P35-DS4 CPU: Intel q6600
RAM: 4 GB Patriot Extreme Video: XFX 7600GT
Case: Antec Take 4 (rackmounted) Drives: 4 Sata II - 1.2 Terabyte
Cooling: Corsair Nautilus w/ D-Tek Fuzion
OS: XP SP2 Audio Engine: Cubase 4.1
Interface: TC Electronic Konnect 24D
Controllers: AlphaTrack, padKontrol, Prodikeys
 

Houndsteeth

Distinguished
Jul 14, 2006
514
3
19,015


1) External radiators tend to use large 120mm fans at fairly low speeds. Sure, I may have 4 120mm fans in a push-pull config on a dual 120mm radiator, but they are all whisper-quiet 100% of the time. Think of water cooling as nothing more than a hyper-efficient air cooling system, where instead of pushing air around to try and pull heat away from various sources, instead we are concentrating the heat at one location (the radiator) so we can use air to remove the heat more effectively. We just happen to use water as our medium to move that heat.

2) Most extreme water coolers usually leave the northbridge, southbridge and RAM to ambient cooling. Some may go ahead and wc the Northbridge, but only because they have to push the voltage in order to get a high CPU oc. Needless to say, with all the tubing for cooling, water coolers tend to be a lot more anal about cable clutter since it restricts airflow even more than it would for just air cooling alone.

3) One of the benefits I enjoy about my water cooled system(s) is that they are whisper quiet, even under load, which is impossible to say about an air cooled box, which sounds like a 747 ready to take off. On top of that, my E6600 at 3.8 Ghz barely even breaks 36°C at load (29°C Idle) in a room that is at a constant 24°C.

4) My main box has been reusing the same watercooling parts for the last three builds. The only thing I have had to purchase have been block adapters for new processor sockets (I haven't even had to buy new blocks) and new tubing (if I have to refit the loop to size due to different mainboard layouts). Even if I had to buy a new CPU or GPU block, the price isn't that much higher than a retail air cooled HSF.

5) Your cousin's fridge PC will run fine for maybe the next year (more than likely less). Eventually, the compressor will wear out because it isn't designed to work with a system that is continually pumping in heat. It works better at maintaining a certain temperature. The only way I can see this as working is if your cousin has another thermal mass (ie, water or something that can absorb heat energy efficiently) so the compressor doesn't have to run hard every time he turns on his PC. In the end, he has a system that is doomed to failure since it is not designed for endurance.