Noctua NH-U12P: Top Performance AND Silence
by Wesley Fink on March 20, 2008 8:00 PM EST- Posted in
- Cases/Cooling/PSUs
Conclusion
Past test results with Noctua coolers taught us to respect Noctua's claims. Like Thermalright, Noctua has a track record of delivering what they promise. However, the two companies really come to cooling with different areas of expertise. Thermalright has a long history of top-performing coolers. Thermalright doesn't ship coolers with fans, so Thermalrights require top-performing fans to end up as top performers. You can also control noise by selecting a fan that exhibits ultra low noise. Fans than can provide good performance with low noise tend to shine on Thermalright heatsinks.
Noctua on the other hand has a sterling reputation for developing the lowest noise coolers around. Their fans are masters of low noise, with respectable performance that is always near the top but rarely challenging for first place performance. For this reason, we were a bit skeptical when Noctua claimed top performance with incredibly low published noise levels. Noctua is capable of some brilliant engineering so we did not dismiss the claim, but we really didn't expect to fully validate their claims in our testing.
We are happy to report that our skepticism was not justified in this case. Noctua really has produced a top-performing cooler with incredibly low noise in the NH-U12P. The combination of staggered Vortex Control notches, SC Drive, and SSO bearings in a nine-blade design does the trick as promised. Paired with the venerable and continually refined Noctua heatsink, performance of the stock Noctua NH-U12P mirrors the top Thermalright all the way to 3.91GHz. Noise remains below our system noise floor throughout the entire range of testing.
The Thermalright goes on to 3.94GHz with a single fan, indicating the design may be slightly more efficient. However, add a second NF-P12 to the Noctua and it also reaches 3.94GHZ and sets new records for cooling temperatures along the way. Even with two fans, noise remains below the noise floor - and you can reduce it even further with shunts if you choose. We saw no reason to do this as two fans at full speed were still as silent as we could measure in our system - with noise levels below a quiet suburban bedroom at night.
For those who don't care about overclocking, the Noctua NH-U12P is still a great choice. It cools as well as the best so far with one fan - at idle and under load at stock speeds. Add a second fan and the stock temperatures are all new records. We haven't tested a cooler that does a better job of cooling at stock speeds. However, the Noctua cooler with two fans at $90 is not cheap, and the Alpine coolers we recently tested do almost as well at about 1/6 the price. For most who won't overclock the Alpines are fine, but if you want the best stock air-cooling you can buy the Noctua NH-U12P with an extra NF-P12 fan is the ultimate.
These test results for the Noctua are truly impressive, but they do not dethrone the Thermalright Ultra-120 eXtreme - at least not yet. The superb component here is the Noctua NF-P12 fan, which is a brilliant piece of engineering. We doubt our Thermalright or any air-cooler will top the 3.94GHz that seems to be the limit of our test bed Core 2 Duo. However, we suspect the Thermalright combined with the NF-P12 could match or even exceed the results with the Noctua heatsink. Having said that the practical reality is it is not easy to mount two NF-P12 fans on a Thermalright, whereas the Noctua heatsink is designed for push-pull and even comes with the extra wires to make it easy.
There is no doubt the Noctua NH-U12P has joined the ranks of the best performing coolers ever tested at AnandTech. With two NF-12P fans in push-pull, it turns in the best cooling results at the highest overclocks we have ever seen. We can also recommend the Noctua NF-P12 fan for performance plus silence on the Thermalright. If you want a complete cooling kit balanced for performance with silence, the NH-U12P is an excellent choice.
51 Comments
View All Comments
Wesley Fink - Friday, March 21, 2008 - link
We covered the Thermalright question head on in the Final Words. The Thermalright is not really set up for push-pull fans so you have to jury rig to get two fans on the cooler. We stated clearly the Thermalright would likely outperform the Noctua with the same fan, but if you wanted to conveniently mount two fans the Noctua was a better choice. Several readers have detailed instructions in the comments on how to mount two fans on the U120 eXtreme with Zip Ties.BTW we have tested every heatsink that was setup for multi-fans with both single and push-pull configurations, so your comment is not fair or accurate. The Thermalright is not really set up for two fans and you end up ripping open the fan wire holders if you try to force two fans. That is why people resort to jury-rigging and Zip ties. We hope Thermalright will address this in the near future as the Thermalright could othwerwise easily mount push-pull fans.
poohbear - Saturday, March 22, 2008 - link
oh ok, i dont recall reading that part of the article, my bad. yea if Thermalright isnt setup for a push/pull config from the factory, it wouldnt make sense to start making customizations just to accomodate it. cheers though.Mr Perfect - Friday, March 21, 2008 - link
At first I was little taken back by the airflow numbers of the fan, but then I noticed they where reporting in Cubic Meters Per Hour instead of the normal Cubic Feet Per Minute. Do you have CFM numbers so this fan can be compared to other fans on the market?Air Flow 92.3/78.5/63.4 cubic meters/hour (1300 stock/1100 LNA/900 ULNA)
poohbear - Friday, March 21, 2008 - link
its 54cfm @ 1300rpm.1 foot = 0.3048 metres; therefore
1 cubic foot = 0.3048*0.3048*0.3048 cubic metres = 0.028316846 cubic metres
and
1 hour = 60 minutes
therefore 1 m3/h = 0.5885778 cfm
hence 92 m3/h * 0.5885778 = 54.1491576 cfm
cheers.
JarredWalton - Friday, March 21, 2008 - link
Some simple math should suffice....1 cubic meter = 35.31467 cubic feet
60 minutes per hour
That gives 54.326/46.203/37.316 CFM, assuming they didn't just put in the wrong abbreviation.
Mr Perfect - Saturday, March 22, 2008 - link
Those numbers make much more sense. Other ~1000RPM 120mm fans also push in the 36ish CFM range.I was hoping Anandtech would put the CFM numbers in the article, it could be misleading if readers don't notice that the rates are not listed in the standard measurement. Anyone who buys a whole bunch of these $21 fans because they think they're putting out 63.4CFM at 900RPM and 12dba is going to be upset.
Baked - Friday, March 21, 2008 - link
If you're not overclocking, Freezer 7 Pro should be the no brainer HSF of choice.Baked - Friday, March 21, 2008 - link
Why bother wasting all that money on trivial performance gain. Just get the Freezer 7 Pro and be done with. Oh wait, I forgot about the epeen factor.poohbear - Saturday, March 22, 2008 - link
freezer pro 7 won't give u a 50%-70% overclock. a 70% overclock is hardly trivial, this is'nt video card overclocking wherein people usually get a 10%-15%.Basilisk - Saturday, March 22, 2008 - link
"freezer pro 7 won't give u a 50%-70% overclock"What am I missing here? As I read the charts in this article, the best listed cooler gave 34% OC on this CPU, while the AFP7 gave 31%. Maybe I'm missing the obvious; or are you reporting un-cited data?
My instinct was the same as others: the AFP7's the better performance/$$ unit, although this new kid has a slight upper end edge.
Not that I don't appreciate the fan improvements. My instinct is there are several further improvements to reduce flow turbulence [noise/efficiency], but we don't see much change in fans.