Sign in with
Sign up | Sign in
Your question

Liquid Cooling Showdown

Last response: in Site Feedback
Share
August 6, 2007 4:03:50 PM

In part 1 of our liquid cooling comparison, Shelton compares the cooling ability of fluids from Thermaltake, SilverStone and Koolance. Part 2 will test the cooling abilities of car anti-freeze and beer.

Watch Part 1: http://www.tomshardware.com/site/flash_videos/liquid_cooling_comparison_part_1__standard_coolants.html

What other fluids do you think we should try next?
August 6, 2007 5:33:53 PM

Please, no bodily fluids, nothing that will cause us to catch on fire and die in a huge burning pile of agony.
August 6, 2007 9:07:18 PM

I wouldn't be surprised if the car anti-freeze would perform the same or a little better than the standard cooling liquids for the PC. Nice thing about the car anti-freeze is that it already contains anti-corrosion agents and cheaper to boot. Question would remain do you use the 50/50 mix of water or pure anti-freeze? I'd be curious to see how that works.

Beer? It does have alcohol but I wouldn't wanna try something like rubbing alcohol as a cooling liquid cuz it's too flammable. Wonder if ammonia would work?

Darkk

Related resources
August 6, 2007 9:17:16 PM

Nice Job THG - quite entertaining!
First - why not more fluids? Second add another pump to add and remove fluids faster - a submersible in distilled for the flush

Second, i am not sure that the system reacted exactly the same each time- why not run orthos or something to get 100% cpu output this would assure a better comparison.

I would expect higher temps as the system warms or the room warms you need readings on the room temp.

You need to test the fluid several times to rule out ambient temp changes and changes in the system temps. If tt fluid is first run it again last, etc.

cool test you need to expand the number of fluids like the over priced xp etc! i saw the beer bottles did it/u get sticky when u drained the beer?

again nice job - looking forward to the next phase!

ok - ya i am pain!

note: water cools the system, the radiator fluid decreases the heat capacitance with increasing levels!

therefore since water cooled computers aren't usually built for sub zero use i expect the radiator fluid to be hotter
August 6, 2007 11:48:22 PM

You state in the video and in the system specs at the end that you're using a Quad Core processor, yet the pics of task manager only show two cores rather than the expected 4.

What's up with that?
August 6, 2007 11:48:53 PM

the video voice says '...of our overclocked QUAD core' processor but the screen shows what appears to be a DUAL core processor.

also, what was it overclocked to? in part II I hope there is a short web page summary with numbers on it.
August 6, 2007 11:56:19 PM

what a waste of time...
This test proves absolutely NOTHING! It only serves as a public mention of the three products.

jeez, lame.

How about measuring ambient temp?
How about running a baseline with pure water?
How about using actual sensory equipment?

Come on ! the coolant has such a little effect on temperature and you are measuring with something as un-sensitive as that thermaltake probe?

Why not use Intel's TAT as to properly load the cores?

also, why not actualy test with various ratios of teh zerex and water...because it will outperform all teh coolants. and so would straight water.

"payed-off'
August 7, 2007 4:39:39 AM

Hey guys, in my water cooled rig, i use FluidXP Extreme, i was wondering if you guys could check out that stuff too, its expensive and i would like to know if its worth the extra money.
August 7, 2007 3:21:06 PM

funnyman06 said:
Hey guys, in my water cooled rig, i use FluidXP Extreme, i was wondering if you guys could check out that stuff too, its expensive and i would like to know if its worth the extra money.

I am also running FluidXP Extreme. I would like to know how this compares as well. It doesn't seem that expensive though, it only cost me about $30 USD and I have enough for the initial fill and enough for an entire refill.
August 7, 2007 5:52:53 PM

ok , fluidXP is a complete ripoff. after some time in th eloop touching metal blocks it no longer is non-conductive. also temps are never better than pure distilled water+bicide.
August 7, 2007 7:19:22 PM

Senater_Cache said:
ok , fluidXP is a complete ripoff. after some time in th eloop touching metal blocks it no longer is non-conductive. also temps are never better than pure distilled water+bicide.

Whatever you said doesn't matter, that's not what I'm asking.
August 8, 2007 12:10:49 AM

Darkk said:
I wouldn't be surprised if the car anti-freeze would perform the same or a little better than the standard cooling liquids for the PC. Nice thing about the car anti-freeze is that it already contains anti-corrosion agents and cheaper to boot. Question would remain do you use the 50/50 mix of water or pure anti-freeze? I'd be curious to see how that works.

Beer? It does have alcohol but I wouldn't wanna try something like rubbing alcohol as a cooling liquid cuz it's too flammable. Wonder if ammonia would work?

Darkk


Since the second video isn't up yet, I won't get too detailed here, but I will say that I'm open to further experimentation with the Antifreeze as well as other fluids!

btw, Shazzbot = Shelton
August 8, 2007 12:23:29 AM

dragonsprayer said:
Nice Job THG - quite entertaining!
First - why not more fluids? Second add another pump to add and remove fluids faster - a submersible in distilled for the flush


The main reasons for not using more fluids had to do with cost and time. We wanted to see how the project would be received by you guys, and personally, I had hoped to get some level of participation out of our readers/viewers in the form of suggestions for fluids/setups and general feedback, both of which I'm very glad to see here already.

dragonsprayer said:

Second, i am not sure that the system reacted exactly the same each time- why not run orthos or something to get 100% cpu output this would assure a better comparison.


This time through we used the burn-in feature of SiSoft Sandra 2005 to loop and attain our work load. Again this was more of a preliminary test than a truely finished, polished project, so I'm reviewing suggestions and discussing methods as well as the lessons learned. Definetly our method of bringing the CPU to maximum load is under review along with the rest. I have a sneaking suspicion that I'll need to play with a couple programs to find the best fit for the projects needs.

dragonsprayer said:

I would expect higher temps as the system warms or the room warms you need readings on the room temp.

You need to test the fluid several times to rule out ambient temp changes and changes in the system temps. If tt fluid is first run it again last, etc.

cool test you need to expand the number of fluids like the over priced xp etc! i saw the beer bottles did it/u get sticky when u drained the beer?


I plan on taking these environmental variabls into consideration should we decide to move forward with more videos and more fluids, possibly with not just multiple test runs per fluid but also with longer test runs. I'm confidant that we'll hammer out these details in short order and be able to have a lot more fun on this!

And unfortunately, yes, yes I did get sticky with the beer >_<
August 8, 2007 12:48:10 AM

phonoflux said:
You state in the video and in the system specs at the end that you're using a Quad Core processor, yet the pics of task manager only show two cores rather than the expected 4.

What's up with that?


Argh! I'm a little peeved at myself for allowing a mistake like that to get through, but thanks for pointing it out, we're making the necessary adjustments so that everything is accurately reported!
August 8, 2007 1:16:10 AM

shazzbot said:
...I had hoped to get some level of participation out of our readers/viewers in the form of suggestions for fluids/setups and general feedback...

-What about FluidXP Extreme?
-What about Corsair Nautilus 500?

It'd be interesting to see how Corsair's only cooling product (that I know of) compares with what you got. :) 
August 8, 2007 2:43:48 AM

shazzbot said:
The main reasons for not using more fluids had to do with cost and time. We wanted to see how the project would be received by you guys, and personally, I had hoped to get some level of participation out of our readers/viewers in the form of suggestions for fluids/setups and general feedback, both of which I'm very glad to see here already.


You need to use Distilled water without adative and with about a 10% Zerex. This setup will both benchmarck with a base line and see if any coolent as good as H2O.
August 8, 2007 3:06:52 AM

so why Molson Canadian for the next test, i think it would be too good to waste on testing the cooling properties, i personally would drink it and get a crappy beer such as laker, or budwiser for the test
August 8, 2007 3:43:22 AM

Senater_Cache said:

How about measuring ambient temp?
How about running a baseline with pure water?
How about using actual sensory equipment?

Come on ! the coolant has such a little effect on temperature and you are measuring with something as un-sensitive as that thermaltake probe?

Quoted for support.

You need REAL temperature measuring equipment(something with .02+ C resolution or better and a plotted precision drift relative to temperature) and REAL procedures(think science lab).

Going at it ad hoc serve no purpose except for product placement.

Volumes have been written on this subject, try and research the work of Bill Adams(BillA) and Stew (Cathar).(procooling.com, xtremesystems.org other places...)

I would assume that there was one or two people in the same room as the tests were being performed.

This alone would invalidate your results.

If you want to make a serious attempt at this then wonderful!

Otherwise this series will be at best useless and at worst(and more likely) misinformation.
August 8, 2007 8:07:28 PM

BGP_Spook said:
Quoted for support.

You need REAL temperature measuring equipment(something with .02+ C resolution or better and a plotted precision drift relative to temperature) and REAL procedures(think science lab).

Going at it ad hoc serve no purpose except for product placement.

Volumes have been written on this subject, try and research the work of Bill Adams(BillA) and Stew (Cathar).(procooling.com, xtremesystems.org other places...)

I would assume that there was one or two people in the same room as the tests were being performed.

This alone would invalidate your results.

If you want to make a serious attempt at this then wonderful!

Otherwise this series will be at best useless and at worst(and more likely) misinformation.


Having done the initial runs now, and seeing first hand what doesn't work as well as what does, I do realize that some changes would be greatly beneficial. However I do feel the need to point out that (and perhaps it's my fault for not making it more clear to begin with) that we're not trying to define maximum cooling performance of each liquid; we're simply not equipped to do that as not only would we need better sensory equipment (a simple part of the problem to remedy), but also a sealed, temperature controlled room without anyone inside for a perfectly ideal environment.

Instead our goal is to show the potential differential between the fluids, or in some cases (like the beer) whether or not it can actually be used as a coolant at all. Further, our intention is to show results as close to what you would get should you attempt this at home, which by default means room temp fluctuations as well as a body or two being present. To that end we're taking the lessons from the first attempt as well as the valuable suggestions and criticism provided by you guys and investigating all possibilities for any future runs!

In short, yes, we want to make a serious attempt at providing something fun and informative!
August 8, 2007 8:23:14 PM

I would be interested in seeing how some beverages with higher alcohol content than beer would perform (vodka, everclear, etc). Since alcohol has a lower boiling point than water (therefore can move heat faster) it doesn't surprise me that beer performed well since it probably bordered on a phase-change cooling solution.
August 8, 2007 8:23:57 PM

martin_man said:
so why Molson Canadian for the next test, i think it would be too good to waste on testing the cooling properties, i personally would drink it and get a crappy beer such as laker, or budwiser for the test


tbh, it wasn't initially the plan to use Molson. We simply were talking about beer in general. Then a couple days later i was at a grocery store and was suprised to spot a box of the stuff (I'm from Vancouver, you see) and picked it up out of nostalgia. Trust me, the rest of that box was enjoyed (responsibly) at home after we no longer needed it for the testing. =Þ
August 8, 2007 8:55:53 PM

I think when you try to use a liquor like 40% vodka instead of the 5% Molson, you'll run into flammability issues. Perhaps a cheap white wine? Usually around 13.5%, not carbonated, isn't thick like some of the red wines can be.

As I recall, ethanol has a much higher heat capacity than water, but a boiling point of 172.4F (78C). You won't have a phase-change coolant without also having a really bad cooling system.
August 8, 2007 10:52:52 PM

As a homebrewer of beer, not electronic devices, I have learned a thing or two about cooling five gallons of boiling hot liquid. The method I use is to emerse a coil of copper tubing into the very hot liquid and then running cold tap water through it. I learned a while back that running the tap water through the coil at full blast does not cool the hot liquid the fastest. The cold water running through the coil has to have enough in the coil to heat up. I imagine that if you were able to vary the speed of the recirculation pump you are using in your experiments that some of the unusual liquids you have been trying could provide better cooling than the comercial products.
August 8, 2007 11:40:38 PM

Anyway... beer seems to work! You better explore that a bit further, otherwise I'll be using a pack of Budweiser next week...
August 9, 2007 12:38:07 AM

car coolant has worse thermo conductive properties than straight water. The reason its used in cars is cause it doesn't freeze at as high of a temp and takes more heat to boil, thats why you do't see straight antifreeze in cars either.
August 9, 2007 12:53:08 AM

battousai831 said:
car coolant has worse thermo conductive properties than straight water. The reason its used in cars is cause it doesn't freeze at as high of a temp and takes more heat to boil, thats why you do't see straight antifreeze in cars either.

Well put. You must not understand what antifreeze is if you think it's used to aid in cooling.
August 9, 2007 1:52:17 AM

gwolfman said:
Well put. You must not understand what antifreeze is if you think it's used to aid in cooling.


Antifreeze was in there because I had previously heard roumors of such attempts, but seen nothing quantative on review sites. This made me curious. If, however, I had heard instead of soysauce being utilized for cooling, then the we would have used that instead :lol: 

But, as previously stated, I'm definetly open to trying different mixtures of water/antifreeze to see the results we get there. There's a lot to sort through before we take the next step, but I'm confident that it will be worth while.
August 9, 2007 2:20:50 AM

I can tell you that putting anti-freeze into a Koolance system will dissolve the seals, crack the polycarbonate heatsinks, and void the warranty. Doh!

It seemed to hold the temps as good or better than the Koolance liquid though. :) 
August 9, 2007 5:23:40 AM

Darklife41, I think you are full of crap, heatsinks are not polycarb and antifreeze does not attack them.
August 9, 2007 4:48:26 PM

Ok I think I can straighten a few things out. But first I think it would be helpful for everyone to know some definitions,

Specific heat - The amount of heat, measured in calories, required to raise the temperature of one gram of a substance by one Celsius degree.

Heat - is the transfer of kinetic energy from one medium or object to another or from an energy source to a medium or object. Such energy transfer can occur in three ways: radiation, conduction, and convection.

The specific heat of water is 1 and that%u2019s about as high of a specific heat as you can get when you consider everyday commonly available liquids. I think the stone gypsum has a specific heat of just above one but that%u2019s in its solid state, I digress back on topic.

So basically water can carry a lot of kinetic energy or heat. The fastest and most efficient way to transfer heat is through conduction (objects physically touching one another to facilitate heat transfer). This is why water blocks generally have a lot of little passageways for the water to travel through because it increases the surface area that the water can touch and there for the water is able to gather up and store as much heat as possible to be carried away (and why HSF have a lot of fins for the air to move over and cool).

Now to explain a few things that people have talked about.

1. Alcohols of any sort won%u2019t work as a coolant very well because its specific heat is not greater than waters. The reason beer worked decently is because it%u2019s mostly water 95% to 92% depending on the beer. If you%u2019re thinking, %u201CWell rubbing alcohol feels cool on my skin.%u201D That%u2019s because it is evaporating off your skin. We can%u2019t have coolant in our pc evaporating on us that quickly or we would need gallons of it per day.

2. The reason antifreeze didn%u2019t work well is because its purpose is like its name suggests, to prevent freezing. It is supposed to be an additive to water (H2O) in your car to prevent the water from freezing in winter and cracking your engine block. Yes it works as a coolant in summer too but that%u2019s why you mix antifreeze 50/50 with water because that mix won%u2019t freeze unless it gets very cold but it works well to cool your engine.

Edit: I think I should say that if you live in an area that never sees cold weather you still need to put antifreeze in your car. It also raises the boiling point of water and acts as a lubricant for your water pump so don't go thinking you can just put water in your car because thats not good for it either.

I%u2019d like to say that what I have said hasn%u2019t been fact checked to an extensive amount. Physics class was a little while ago and I typed this up in a bit of a hurry, but it is for the most part accurate. I very well could have made a mistake and welcome corrections that need to be made if I or someone else sees them.

August 9, 2007 6:39:48 PM

shazzbot said:
Antifreeze was in there because I had previously heard roumors of such attempts, but seen nothing quantative on review sites. This made me curious. If, however, I had heard instead of soysauce being utilized for cooling, then the we would have used that instead :lol: 

But, as previously stated, I'm definetly open to trying different mixtures of water/antifreeze to see the results we get there. There's a lot to sort through before we take the next step, but I'm confident that it will be worth while.



I don't think you get it about Anti-freeze. Anti-freeze itself has a significantly lower capacity to absorb heat. Adding in any faction of Anti-freeze to water will lower the cooling capacity period. It's only functions are to significantly enhance the range in which water stays liquid, and to prevent corrosion in Aluminum/Iron cooling systems.

The only possible way it could be useful is if you were to stick the radiator in a freezer and didn't want your water to freeze when the pump is shut off.

That said, it was worth a single attempt just to show how pointless it is to use in a conventional room temp environment.

Automotive products that are worth tring are Redline's "Water-Wetter".

samir_nayanajaad said:
Edit: I think I should say that if you live in an area that never sees cold weather you still need to put antifreeze in your car. It also raises the boiling point of water and acts as a lubricant for your water pump so don't go thinking you can just put water in your car because thats not good for it either.


In no way that I know of could Anti-freeze be considered a lubricant for your water-pump. The bearings in every water-pump I've even seen (or replaced) are completely sealed. Only when these seals are breached (by wear or corrosion) and the bearings are actually exposed directly to the coolant do the water pumps fail, and every one I've ever had to replace failed due to bearing seal breach.

Plain water in your engine is fine under many circumstances. In certain engines, however, it will absolutely create corrosion issues, filling your cooling system with rust particulate, eventually clogging either your t-stat or your heater core. This is very common in certain dodge/chrysler engines, even with proper use of anti-freeze because of poor choice in materials.
August 9, 2007 8:24:09 PM

I agree, you won't be able to improve upon water. Basically, you need density * specific heat capacity / molar mass > 4.18. Then your liquid is better than water. specific heat capacity is measured in J/mol*K (only chemists use calories). I saw one website claim ammonia is better, but it boils at -30C, so we couldn't use it, even if we thought it would work. But, it doesn't fit the equation, since density = 0.682g/mL, specific heat capacity = 80.08 J/mol*K and molar mass is 17.03g/mol.
August 10, 2007 3:58:20 AM

lol! nice part 2!

not surprising, seeing as beer is ~95% water, I am sure with time you will get some allege build up?

long term: corrosion probably is not an issue

i think will we hook the keg to the system with flat bear then output to liquid chiller after the cpu,gpu and chipset we can then finally add CO2 to system and output the coolant directly to mugsl

geek bar!

o ya the, sorry,sad plug:


August 10, 2007 9:16:35 AM

RoyalCrown,
You ever used a Koolance water block? Apparently not, or you'd know that they are injection molded plastic surrounding the metal inserts, and the coolant passes through the plastic. Crack the plastic and they leak.

Koolance told me that the only time they'd heard of that happening was with someone else's coolant. Within a month of changing to anti-freeze the first block cracked. The 2nd one lasted about the same. The seals in the 18 month old Exos2 went out at the same time as the 2nd CPU cooler (that's actually when I noticed it), which was the end of water cooling for me. The leaking seals took out a PX7900GTX TDH Extreme, and very lucky it didn't take out more. Now I don't know for sure if the seals were attacked by the anti-freeze, but it's funny that they went at the same time. The seals should have lasted much longer.

My best guess is that this is polycarbonate, as I was an injection molding supervisor/molding technician for 17 years. These are definitely injection molded inserts as I can see the ejector pin marks, flow lines and gates. I do know that polycarbonate is attacked by many chemicals. That's how we used to get stuck parts off molds, apply a little acetone and PC shatters like glass. That's exactly what these CPU blocks look like, disintegrated at the base and cracked all the way around.

That's also where I learned about cooling with water. The most important variable is turbulent flow. I have yet to see a pump for computer systems attain turbulent flow, which is about 5 gallons per minute of circulation. With proper flow, heat can transfer in and out of water faster.

My thinking with the anti-freeze is that its much more dense, thus would take longer to heat up, which should be good for short term overclocks. Unfortunately, it would also be slower to cool back down. At any rate, I won't be using it again.

Here's a couple pics. Sorry for the quality but my camera sux for this kinda stuff. The half that didn't come off is also cracked all the way around and the entire bottom is brittle. :) 

http://www.ultramaxcc.com.au/images/cracked1.jpg

http://www.ultramaxcc. com.au/images/cracked2.jpg
August 10, 2007 9:22:13 AM

I can't fix that bottom link, but just take out the space and it'll work.
August 10, 2007 6:56:05 PM

Dont know if anyone said this yet, but about the beer will obviously cool better than a 25% anti freeze solution. anti freeze only prevents freezing and helps stop corrosion it is not good for cooling. Ethylene glycol is a MUCH worse conductor of heat than water, you cannot beat water ITO cooling if it was not for hydrogen bonding there would be no life on earth. 4200J/l/K vesus whatever else, and since everything else is lower than 4.2 kj good luck cooling something with anything other than water and expecting results that are as good.

Beer will be VERY crappy to clean cause it contains a lot of malt and still quite a bit of sugar, you will therefore get a lot of Maillard reactions when your sugars react with your proteins and form a nice sticky brown mass all over you cooler and in your tubes. Also the etOH will not be a good thing either. They make beer out of whatever is the cheapest crap they can find for their particular yeast to grow on, so you will be putting all kinds of horrible stuff in your pc, let alone your mouth which, it seems, is attached to something dumb enough to put it in a pc and down it's throat and not the drain where it belongs.
August 10, 2007 8:07:19 PM

BTW, Koolance's water blocks may be acryllic. Can't think of what other material they could possibly be besides PC. I'm still pretty sure the anti-freeze is what degraded the plastic though.
August 11, 2007 4:47:28 PM

I watched both videos Aug. 11th.

Part one showed:
liq. idle temp liq. load temp CPU idle CPU load
Thermaltake 26 29.4-29.6 21 43
Silverstone 26.4 28.6 21 48
Koolance 25.5 29.9 22 49

Thus, in the version of the video I watched 6 times (while taking notes); the Thermaltake liquid is shown to be by far the best at cooling the CPU under load. The Thermaltake liquid keeps the CPU under load five (5) degrees Centigrade cooler than the Silverstone liquid!
August 17, 2007 6:56:08 PM

Darklife, did you get any pictures?

Maybe we need to test more for endurance and corrosion.

One thing is for certain. There's not enough hard, tested data in this area. What fluids cool the best? How often do you really need to change the fluid? Does the really expensive stuff do anything? And there's only one way to get this data.

More tests. Longer and harder tests.

Questions? Comments? Test suggestions?
August 25, 2007 6:35:42 AM

Pictures are the links in my post above. :) 
December 1, 2007 7:16:13 PM

I found it very interesting that anti-freeze was used.... I don't believe that antifreeze has better heat transfer capabilities than distilled water, its primary goal is to lower the freezing temperature of the cooling liquid, and secondly many manufacturers try to include additives to help improve heat transfer capabilities which were lost by adding the anti-freeze components. I would like to see a test using a glycol ether mixture such as waterwatter made by redline, this cooling liquid is used to maximize cooling capabilities in race cars with no freezing temperatures considerations which is more applicable to the current application. The optimized mixture of water and watterwetter should have a low viscosity and pump easly through the standard PC water cooling pump

Chris


!