Finding the Best CPU Cooler: 10 units reviewed and tested

[Guest]

Haha, nice to see the Noctua NH-U14S win that, as it did another heatsink round-up at Tom's Hardware recently. This is cooler I had elected for my latest build and I'll keep heartily recommending it.

 

[Steve]

Solid review nonetheless.
In the future it might help to make it look more professional if you'd include a specific section with a few more details about the testing methodology such as case/other hardware, thermal paste, where/how you measured the sound...

Most people won't read it but it's good documentation.

Thanks. Ironically 'most people' might include yourself

Do you mean something like on page six (titled Testing Notes) which specifies how we tested, hardware, programs and ambient room temperature. The case was also mentioned, Cooler Master HAF XB and how the heatsinks were orientated.

Each cooler was tested with the supplied thermal paste and fan. The only thing we didn't go into great detail about was the sound testing.

 

[Guest]

"For $70 to $80 I would go custom water cooling"
--> unlike you, for $100 to $130 I would go for closed watercooling system

"I dont think they should last 1-2 years then corrode as some have said. Car radiators last about 10 years or longer running at 200+ degrees with 15lbs pressure, they are just made of aluminum and plastic."
--> you can't be serious comparing car radiators (car cooling system) with PC watercooling system.

 

[Jad Chaar]

Yes, the Noctua NH-D14 and Hyper 212 Evo are good coolers, but werent tested. You guys gotta appreciate the work Steve and the TS team did nonetheless!!!

 

[Dukenukemx]

You want quality water cooling components IMO. Plus, WCing a HTPC just isnt necessary.

My main PC has had water cooling for two years, while my HTPC has had it for 1 year. The reason for the water cooling was quiet it down and to add more room. A decent air cooler is super larger, and takes a lot of space. Also any decent air cooler will usually have a large loud fan. With water cooling I was able to gain improved cooling, without the need to take up large amounts of space, and the HTPC is much quieter.

 

[Dukenukemx]

A good water cooler is only good if after 2 years it hasn't accumulated any corrosion or produced any leaking. And most reviews are from people who haven't owned the product long enough to know if it is any good. Most coolers fall victims to corrosion and produce leaking that may easily kill the system. And even if it takes 3 years for one to corrode to that point, I wouldn't want it in my system.

Never seen a water cooling setup leak so far. I've had water cooling setups for nearly 10 years, and they never leak. Water pump failure yes. Pinched tubing yes. Clogged tubing yes. The only place it could leak from is the radiator, and I've yet to see it happen. If you're really anal, you could keep a decent amount of the components outside of the case, like the radiator and pump.

There's also some water blocks that have rubber seals, that allow you to take it apart to clean them. Haven't seen them leak yet either. The only real problem with water cooling is that every so often you want to empty out the coolant, clean the system, and refill with new coolant. Every 6-7 months this needs to be done.

And why bother, Noctua NH-D14 is just as quiet as any water cooler, it makes zero noise.

The Noctua NH-D14 is also very very large, and dumps all the heat in it's surroundings. Where as a water cooling setup will remove the heat directly outside the case. It's also listed as $95 on Newegg, which is very expensive for an air cooler.

With a water cooling setup I can and do cool the GPU as well, at least in my main PC. For additional cooling, I can either add another radiator without taking up a boat load of room, or just buy a water pump with higher flow.

 

[cliffordcooley]

Also any decent air cooler will usually have a large loud fan.

I think you have that backward, as the larger the fan the less RPM's needed to move air. It is the smaller fans that are the noisiest. Besides those large fans you speak of are also needed on water cooling setups.

 

[Dukenukemx]

I think you have that backward, as the larger the fan the less RPM's needed to move air. It is the smaller fans that are the noisiest. Besides those large fans you speak of are also needed on water cooling setups.

At 120mm, it's all in the fan. At low speed a 120mm fan is nearly silent, but crank them to max speed and they get noisy. To reach the efficiency of a water cooler, those air coolers need the fans at max.

Also take into account that since air coolers dump air inside the case, you need a lot more case fans to remove that hot air. Technically speaking my case fan is my water cooling fan, cause the radiator is placed where a 120mm fan would be. The fan just sits on top of it. So generally a water cooling setup uses less fans then air cooling, and uses them at lower speed. Since my main PC uses a water block on the GPU, that's also one less fan.

 

[dexterkarthik]

Could you please recheck FRIO OCK?
I am running Intel Core2Quad Q6600 at stock 2.4GHz in an avg ambient room temp of 30-33 with idle temps between 31-33C and Max temp while playing Crysis 2 and 3 is 54C with fan at stock speed of around 1200-1400 rpm

 

[Dukenukemx]

Could you please recheck FRIO OCK?

I am running Intel Core2Quad Q6600 at stock 2.4GHz in an avg ambient room temp of 30-33 with idle temps between 31-33C and Max temp while playing Crysis 2 and 3 is 54C with fan at stock speed of around 1200-1400 rpm

If you really wanna know what max temp your PC can handle, then get Prime95 and run the "In-Place Large FFT's" test, which creates a boat load of heat. Nothing will get your CPU hotter then that test, but it's also a very dangerous test. Crysis 3 is pretty good CPU test, but not as good as Prime 95. BTW, Prime95 on my HTPC with AMD Phenom II X4 820 2.8Ghz gets 42C with 10 minutes of Prime95 on that setting, while typing this. Which is pretty good for a cheap water cooling setup.

 

[GhostRyder]

I had a Frio CPU fan cooler for one of my previous processors and it was pitiful to be honest. I really like the Noctuas a lot better because they actually feel like decent improvements. The only advantage I can say the Frio has, is its price.

 

[dexterkarthik]

If you really wanna know what max temp your PC can handle, then get Prime95 and run the "In-Place Large FFT's" test, which creates a boat load of heat. Nothing will get your CPU hotter then that test, but it's also a very dangerous test. Crysis 3 is pretty good CPU test, but not as good as Prime 95. BTW, Prime95 on my HTPC with AMD Phenom II X4 820 2.8Ghz gets 42C with 10 minutes of Prime95 on that setting, while typing this. Which is pretty good for a cheap water cooling setup.

Yes, I am aware of that and also furmark for GPU's, just don't want to strain my hardware - want to extract the maximum life out of it, so only I have not even OC'ed my CPU yet. Being serving well so far for my needs. The next upgrade will only be a SSD.

 

[Steve]

Could you please recheck FRIO OCK?
I am running Intel Core2Quad Q6600 at stock 2.4GHz in an avg ambient room temp of 30-33 with idle temps between 31-33C and Max temp while playing Crysis 2 and 3 is 54C with fan at stock speed of around 1200-1400 rpm

Are you suggesting that because you have a 95 watt processor that’s not overclocked our results are wrong? As the testing notes pointed out each cooler was tested 3 times anyway which included being re-seated each time. I would question your own results as there is no way these air-coolers work at ambient room temperature, the computer would have to be turned off :S

You say your system runs at 54 degrees under load, that’s 10% cooler than the results we got. Yet our processor has a 37% higher thermal design power rating.

 

[St1ckM4n]

Comments are hilarious as always. Steve must emit an aura of "hate on me, everyone!!".

I have NH-U12P SE2, and I purchased it based on the fact that an $80 air cooler out-performed $120+ closed-loop systems and was quieter.

My HTPC has another Noctua in it with a 140mm fan. I'd like to know how a water cooling setup with fan + pump can be quieter than me, with ULNA.

 

[Steve]

Comments are hilarious as always. Steve must emit an aura of "hate on me, everyone!!".

I have NH-U12P SE2, and I purchased it based on the fact that an $80 air cooler out-performed $120+ closed-loop systems and was quieter.

My HTPC has another Noctua in it with a 140mm fan. I'd like to know how a water cooling setup with fan + pump can be quieter than me, with ULNA.

So you see it too

 

[dexterkarthik]

Are you suggesting that because you have a 95 watt processor that’s not overclocked our results are wrong? As the testing notes pointed out each cooler was tested 3 times anyway which included being re-seated each time. I would question your own results as there is no way these air-coolers work at ambient room temperature, the computer would have to be turned off :S

You say your system runs at 54 degrees under load, that’s 10% cooler than the results we got. Yet our processor has a 37% higher thermal design power rating.

I never implied that your results are wrong....just felt bad that the cooler I bought (first one [and probably the only one] in India) performed bad in the test while it is performing to my satisfaction in my workstation and hoping some to see it better {http://www.kitguru.net/components/c...-ock-cooler-review-better-than-noctua-d14/5/}...

I was running on stock heat sink for a few days before going for the OCK and the idle temps in Indian [Bangalore] Summer (atmospheric temp >35C) were about 42-44 and max went to 70-73.....the same results with OCK were 39 and 55 respectively.....in Winter when outside temps fell to 9-10 the idle temps were about 20-21, this may be because of the poor air circulation in my room with the doors closed most of the time which I would say increases the ambient temp by at least 3-5 deg....hence I calculated the room temp around 30 as the outside temp is about 24 now

my ATI Radeon HD 5870 (stock) idle temp is around 35 as well and max 64 while CRYSIS'ing at 1440*900 High Settings [with custom fan profile in MSI Afterburner]

PS - Core 2 Quad Q6600 TDP is 105W I guess.

 

[Steve]

Okay so you were just telling us to recheck the FrioOCK results on the basis that you like it. We never said the FrioOCK wasn't a drastic improvement over the Intel box cooler, especially the LGA775 box cooler.

The Q6600 was improved with a newer stepping which reduced the TDP to 95 watt, either way even at 105w my original point still stands. That being you are comparing a 95 or 105w processor to a 130w processor, that and you are comparing the FrioOCK to a stock Intel box cooler and not something like the Noctua NH-U14S :S

Also that link that you provided didn't show the difference between the Thermaltake FrioOCK and the Noctua D14 in terms of noise. The FrioOCK is extremely loud which is why we limited it to 1300 RPM, even then it was noisy. I assume our results are comparable to their medium. In no way is the FrioOCK as good as the Noctua D14, other than it’s a lot cheaper of course.

Direct quote from the review you linked...

"[FONT=Arial]Thermaltake have been highlighting an apparent 7c benefit over the class leading Noctua NH D14. We can categorically state that this is [/FONT][FONT=Arial]utter nonsense[/FONT][FONT=Arial]."[/FONT]

 

[dexterkarthik]

I was just astonished in the difference between OCK and NH-U14S considering NH-D14 performs better than U14S...hence the doubt

best way to clear the air would be to include a NH-D14 in the results (also Hyper 212 Evo, the budget performer super-king) if possible

 

[dividebyzero]

I never implied that your results are wrong....just felt bad that the cooler I bought (first one [and probably the only one] in India) performed bad in the test while it is performing to my satisfaction in my workstation and hoping some to see it better

There are too many variables to take into account to compare a single cooler in isolation against a completely different system. Heatspreader size and underlying die layout are going to give variances in results*. Assuming that ambient conditions are equal (which they wont be - temp, humidity, chassis cooling, general airflow, heat map of the socket area) you would still need other heatsinks to test in order to gauge their effectiveness on the same system, since the LGA 775 (784mm²) and LGA 2011 (1440mm²) are fundamentally different in too many ways

* As an example, Frosty Tech tests with both an AMD (125W) system and an Intel (150W) system. If their case the lower wattage AMD system produces a higher temperature delta over ambient than the Intel system. In part, this instance AMD and Intel measure TDP quite differently. A direct comparison taking into account the turbo feature of the IB-E might be this one from Hardware France which shows the disparity between the Q6600 and the 3960X in actual power usage

 

[Guest]

Yes but only on the Archon SB-E X2. Maybe it's just an optical illusion but those pictures appear to show the fans pulling from the back and pushing towards the front of the mobo.

 

[Steve]

Yes but only on the Archon SB-E X2. Maybe it's just an optical illusion but those pictures appear to show the fans pulling from the back and pushing towards the front of the mobo.

None of those coolers were tested on that board, it was used purely for demonstration purposes I.e. photos. The motherboard used for testing was the Asrock X79 Extreme11 and that stayed installed in the Cooler Master HAF XB, it would have been a real chore removing it each time for the photos so we streamlined the process with the Gigabyte X79S-UP5 WiFi.

 

[Guest]

Each cooler was tested with the supplied thermal paste and fan. The only thing we didn't go into great detail about was the sound testing.

So no consistent 1 paste was used? A variety of pastes were used? I could see this skew the data a little bit. Is there a review available on the different pastes?

 

[cliffordcooley]

So no consistent 1 paste was used? A variety of pastes were used? I could see this skew the data a little bit. Is there a review available on the different pastes?

I agree, the best cooler is still unknown. I still don't know which cooler is best, when different fans and thermal compound is used. Sure we have a good idea which is best using the supplied fan and compound. But to be honest that doesn't tell the whole story.

 

[dividebyzero]

So no consistent 1 paste was used? A variety of pastes were used? I could see this skew the data a little bit. Is there a review available on the different pastes?

Using a variety of pastes would also skew the data. The review is a "run what you brung" comparison. If a particular cooler is shipped with some generic goop (*cough*Xigmatek*cough) in a sachet in order to keep the price low, then adding an aftermarket TIM syringe to the cost inflates the price....a price that is part of the scoring system of the article. If the object of the exercise was to test the best heatsink design rather than the retail cooler package, then I could see the validity in using an aftermarket TIM like IC7 (as well as a couple of Sanyo SanAce 150 cfm fans).
For the record, most quality thermal pastes are within a couple of °C of each other when used as a CPU TIM ( high-end GPUs tend to show a larger delta). As an example, the Noctua coolers use NT-H1 and the Thermalright coolers use Chill Factor III. You can see from Skinneelabs testing that the TIMs exhibit similar performance

I agree, the best cooler is still unknown. I still don't know which cooler is best, when different fans and thermal compound is used. Sure we have a good idea which is best using the supplied fan and compound. But to be honest that doesn't tell the whole story.

That becomes a complex testing criteria, especially if you factor in absolute cooling ability versus produced noise since both have an impact of the end usage profile.
Xbit labs test with a single TIM (MX-4), using a variety of aftermarket fans at different rotational speeds (bear in mind that the two Corsair AF140's are an additional $38)

 

[Steve]

I agree, the best cooler is still unknown. I still don't know which cooler is best, when different fans and thermal compound is used. Sure we have a good idea which is best using the supplied fan and compound. But to be honest that doesn't tell the whole story.

You are using the term "cooler" not heatsink. The cooler includes the heatsink, fan(s) and thermal paste, its a kit and that's what we tested, the kit! Why would you buy an expensive cooler, then remove the fans and throw away the thermal paste? If you were going to do that you would just buy a heatsink like the Prolimatech Megahalems and add your own gear.