Improving system cooling has been a basic part of improving system performance since overclocking began. To squeeze more performance out of computer components users found ways to run them faster than specified. When they hit walls that prevented going even faster they found ways to better cool the components. The better cooling allowed for running the system even faster until another wall was encountered.

The quest for better cooling has been very creative, with solutions as simple as added fans, progressing to larger fans and heatsinks, peltier (thermoelectric cooling), water cooling, and finally the king of cooling - phase change. There are also the truly "benchmark only" cooling solutions like liquid nitrogen cooling. This is certainly not usable in a normal computing environment, but it is a favorite with competitive benchmarkers - who seem to live for the top of the ORB or the highest benchmark scores. This type of cooling is mainly built by the benchmarker and exists for that fleeting benchmark record that will be broken tomorrow by another benchmarker.

As cooling complexity goes up, so does the associated cost. At the extreme end you will find some top phase change units selling for $2,000 or more. You have to be a very serious overclocker or benchmarking enthusiast to invest that kind of money in cooling. Fortunately, you can achieve significant improvements in system cooling and overclocking for a lot less money. The sophistication of cheaper air and water cooling solutions has improved in recent years, and there have also been genuine efforts by the cooling industry to deliver better cooling performance at a lower price point.


The Tuniq Tower 120 is one example of this. As imposing as the multi-finned, heat-piped Tuniq looks it is worth keeping in mind that the total cost of the Tuniq kit is only about $50. The question we will try to answer in this review is whether the Tuniq is effective at improving CPU cooling? If so, does the improved cooling allow higher overclocks than a stock cooling solution? Finally, the impact on system noise will be measured.

Does Cooling Matter With Core 2 and AM2?
Comments Locked

50 Comments

View All Comments

  • monsoon - Monday, January 15, 2007 - link

    Hello,

    I was a little surprised not tp see comparison benchmarks to a TUNIQ + E4300. I would love to see how much higher can the little buddy be pushed to a overclock with a TUNIQ.

    Also, how comes lower clocked CPUs are those that can get higher ? ( sorry, I'm a newbie to PC modding )

    THANKS
  • Great Googly Moogly - Monday, January 15, 2007 - link

    Do not use the word "silent"; use "quiet". Also, you'd do better if you at least gave some impressions of the quality of noise, as that is just as, if not more, important.

    The rest is good.
  • RobbieMc - Monday, January 15, 2007 - link

    I recently purchased a Scythe Ninja to cool a QX6700. I was considering purchasing the Tuniq Tower, or the Ninja, and was told the ninja had slightly better cooling. After installing it (properly), I found that the Ninja ran nearly 30 degrees F hotter than the stock cooling. I was under the impression that the Ninja and the Tuniq Tower had about the same performance, but based on this review, if my data is right, then the Tuniq would be much better. I'm wondering what the weak link is with the ninja. Is it really possible that the stock cooling is substantially better? Do you all suspect I had an improper installation (I am convinced I did everything properly), is the Tuniq better than the Ninja, are my results at all normal? Thanks.
  • xsilver - Monday, January 15, 2007 - link

    http://www.madshrimps.be/?action=getarticle&nu...">http://www.madshrimps.be/?action=getart...ber=2&am...

    a non obvious installation problem is too much thermal grease
    either that or malfunctioning probes?
  • RobbieMc - Monday, January 15, 2007 - link

    The first time installing, I did put on WAY too much thermal grease, but I soon realized this was bad, and reinstalled it with a very thin layer of Arctic Silver 5 grease. I don't think the probes are bad either because the temperature readings I'm getting are within 1-2 degrees of Anandtech's stock cooler temperatures (131F under load).
  • Shark Tek - Monday, January 15, 2007 - link

    Those stock temps are quite high if I compare them to my 939 AMD 4400+ X2 setup. At 2.8GHz I run 35'C idle and 50'C load.

    Which is the maximum or limit temp that a Core Duo 2 can stand in such way that it will work flawlessly? I mean, in which point you need to worry about temperature?
  • Wesley Fink - Monday, January 15, 2007 - link

    How are you measuring temperatures? Most motherboards are notoriously inaccurate in CPU temp measurements.
  • Zaitsev - Monday, January 15, 2007 - link

    Great review, Wesley. I'm so happy that Anandtech is finally testing cooling solutions again. It would be nice if you could comment on the weight of the heavier coolers in the next article, since that is a common complaint about heatsinks like the tuniq tower or infinity. Look forward to the rest of the series.
  • mobutu - Monday, January 15, 2007 - link

    I suggest that Anandtech take a look at http://www.silentpcreview.com/">http://www.silentpcreview.com/ and learn how to do a "noise" test. If 34db is quiet then how about 22-25db? Definitevly 34db is LOUD.

    Otherwise pretty good review, but to compare aplle with aplle then you should test at least Thermalright Ultra 120 and Scythe Ninja (not to mention similar solutions from zalman, thermaltake and the others).

    Cheers!
  • Wesley Fink - Monday, January 15, 2007 - link

    We are very aware of some of the excellent "noise level" testing done by Silent PC. The information is very useful for those who are looking for the lowest possible noise.

    However, all our benchmarking tries to factor in the "real world" where users are operating their computers. The real world in this case is a "subjectively" very quiet 520W OCZ Power Supply at a noise floor of 38.3db and our test room at 36.3 db with all computers and fans turned off. To mainteain noise in the mid-30's we have to leave off all other computers, ac, and heat in the lab during our sound measurements. At those db levels, performance at 24db is interesting, but it doesn't tell us much about the noise of a cooler in a working computer in our test room.

Log in

Don't have an account? Sign up now