Results – maximum speed
While modern processors offer high performance, this is inevitably linked to the need for highly effective cooling. So I’m always happy to welcome a cooler on the market with the promise of high effectiveness at an interesting price. One of them is the AK620, which represents the top of DeepCool’s range. It offers a modern design, an attractive price tag and the potential to cool even high-end processors. So we’ll see how well it performs in our stress tests.
Results – maximum speed
Why is there a missing value in the cooler chart? Either because the cooler could no longer handle the load in the specified mode, or because the fan could not be slowed down to the required level, typically the mode with the lowest noise level.
Continue: Conclusion and evaluation
- Contents
- Key features
- Measurement methodology
- Results – 36 dBA
- Results – 39 dBA
- Results – 42 dBA
- Results – maximum speed
- Conclusion and evaluation
A Gamers Nexus video (https://www.youtube.com/watch?v=UGM6FcLDckk) shows the AK620 outperforming the D15 when noise normalized, but your tests show otherwise (at all tested noise levels). I wonder what could the possible reasons for such discrepancy? Maybe different CPUs, or different location for measuring sound, or some other reason?
Hello. Yes, there are multiple reasons and suspect why result differ. Used CPU, used sound meter, distance to sound meter, angle to sound meter etc. That is why you can never directly compare two differently conducted tests, you would be comparing apples to oranges.
Pavel Sekerka replied too diplomatically. It always makes me sad to see the GN being held up as some kind of standard, because they don’t deserve it. Apart from the faux pas when Steve once “discredited” all those who analyze CPU coolers (and only so he could promote his synthetic load on CPU cooler tests) and yet later switched to in vito methodology and still tests those coolers on Ryzen, which is a rather inappropriate processor for evaluating the performance of coolers, if it is a single test platform, their results don’t really compare to the Kitguru tests. They also use an AMD processor (R9 5900X) and the AK620 only barely beats the NH-D15S at higher noise levels. That means the variant with one fan, which is significantly weaker compared to the two-fan (NH-D15) at the same noise level, and the lower the speed, the more the difference deepens to the detriment of the NH-D15S. I have found tests on Intel platform, which ratios (between AK620 and NH-D15) correspond to ours. Apart from GN, DeepCool’s cooler does not perform better anywhere where there is a hint of some proper work on the test methodology. Don’t get mad at me, but I find anything from the GN guys very untrustworthy.
Their videos are sold by either a negative/ridiculing emotion (which draws people in and they click on it) with strong cherry-picking and if the rating is to be positive, I suspect it won’t be “free”. Noctua, according to their own words, never paid anyone for advertising (their marketing is simply based on a different system), so I’m not surprised that the GN cooler is presented like this.
The Deepcool AK620 has a great heatsink without a doubt, but considering the comparison of the aerodynamic properties of the NF-A15 PWM (NH-D15) and the FK120 (AK620), it is really unlikely that the Noctua cooler would have a worse cooling performance/noise ratio. This is regardless of the processor used and other differences in the test methodology. Personally, I don’t see any special expertise at GN, much less the moral credibility to question other tests based on their performances.
Steve Burke likes to impress people with how many years of experience he has with testing what components. So I’ll also throw my reference in here, as far as work on CPU cooler tests is concerned: https://www.hwcooling.net/en/gigantic-retrotest-of-80-cpu-coolers-2005-2015/
So much from my side and possible reasons why the tests are diverging. For the AK620 with fans with significantly lower airflow to beat the NH-D15, the Noctua heatsink would have to be pretty inefficient. And it isn’t. We have tested it against 80 others with the same, standard fans – the NH-D15 is on the top. It’s from the test I linked to above. We worked on that massive database for a very long time on the now defunct ExtraHardware.cz server.
Thank you for your insights.
I personally found most of GN’s stuff to be alright and in line with many other outlet’s findings, but now that I think more about it, noise testing seem to be the outlier. A 120 mm dual-fan tower surpassing a 140 mm dual-fan tower when Noctua themselves can’t even do it with their own U12A? There’s likely something wrong in the testing process. They also do only one single noise normalized level which the Fuma 2 can’t even reach at full speed, making the results quite pointless to me even if they are valid. In fact my dissatisfaction with noise-performance testing methodologies (not just GN but also most outlets) is what brought me here to hwcooling.
Thank you for your understanding. I was a bit worried that my post would be misunderstood (when someone dares to speak out against the outputs of “big” youtubers, it’s usually, let’s say, not easy :)). Yes, we also test hardware because we haven’t come across anyone who can evaluate it as a whole objectively based on appropriate measurement techniques. Especially when it comes to fan tests.
Ľubomír, have you tested the Thermalright Peerless Assassin 120 SE?
Thank you for asking, but I haven’t tested these big coolers for quite a while. And Pavel hasn’t gotten to it yet either. We will put it on the “attractive, take a look at it” list though. 🙂
Thanks for the great work!
Greetings from Brazil!
One thing I’ve been missing is sound frequency measurements for cpu cooler reviews. Specifically when installed on the cpu cooler it is delivered with instead of a separate test on a separate radiator.
Or do the different radiator fin densities/designs not affect sound frequencies in a noticeable way?
The different shape of the fins has, of course, an impact on the frequency characteristic of the sound. Therefore, in tests with normalized noise levels, the Fera 5’s heatsink in certain settings achieves lower cooling performance with NF-A12x25 (than with Fluctus 120 PWM) and at the same time Fluctus 120 PWM may not achieve better results on the Noctua NF-A12x25 heatsink. I don’t know if this is exactly the case, don’t take this particular comparison literally, but just as an example. My colleague and I (the author of the CPU cooler tests) will deal with this issue in the framework of the thematic articles supported by the tests.
In the case of high-end coolers, the manufacturers fine-tune the shape of the heatsink with respect to the fan characteristics and with an eye on the quietest possible operation. So yes, you are right and thank you for this comment. Maybe it will give Pavel a little kick to start using the microphone he has had for more than a year for these purposes. 🙂