Temps, Noise, and Conclusion
Test Setup and Methodology
Test Platform | |
---|---|
Processor | Intel Core i5-6600K |
Motherboard | ASUS MAXIMUS VIII GENE (mATX Intel Z170) |
Memory | Crucial Ballistix Sport 8 GB 2400 MHz DDR4 |
Graphics Card | XFX AMD Radeon 5450 (Fanless) |
Storage | OCZ Vertex 460 120GB SSD |
Cooling | EK-XLC Predator 240, Corsair H100i GTX, Corsair H75, DEEPCOOL Gabriel, Noctua NH-U9B SE2, Noctua NH-D14 (PWM), Scythe Ninja 4 |
Power Supply | Corsair TX 650W PSU |
OS | Windows 8.1 64-bit |
The test setup I employ for cooling is very simple, with all components tested under the same conditions on an open test bench. To provide accurate noise readings a passively-cooled graphics card is used, and the power supply's fan does not spin under the loads I'm using for these tests. Temperatures were recorded using RealTemp software, with the hottest core at the end of one of the identical 5-minute long tests used for these results.
As we are limited to whole numbers with monitoring software such as RealTemp, it seems safe to assume that I might have up to a +/- 1 ºC margin for error using this method. Room temperature was more precise, with a reading of ambient air (in 0.1 degree increments) taken with each CPU reading, and the final number adjusted to reflect the CPU temperature above that current room temp.
Noise was measured using a digital sound pressure meter positioned exactly 24 inches from the edge of the test system's motherboard. (Note that while great care was taken to take accurate readings, with multiple samples taken for each result, the limitation of my instrument is an accuracy of +/- 1.5 dBA.) The same meter was used for all tests with a noise floor of 33.4 dBA.
Temperature Results
With the test system's Intel Core i5 at stock speed the Ninja 4 (with fans at high speed) was very close to the performance of the Noctua NH-D14 cooler, coming out on top under maximum stress. This was an excellent showing considering the NH-D14 is one of the best performing coolers on the market, besting both the Corsair H100i GTX at its quiet setting, and the Corsair H75.
At the medium and low fan settings the Ninja 4's performance was nearly identical, and actually just a bit better on low than at the medium fan setting; though factoring in margin of error negates this. What these tests demonstrate is that the Ninja 4 is capable of performing as well at the lowest setting as it can at medium (at stock CPU speed, anyhow), so users can opt for the low fan setting (over medium) to reduce noise without impact.
Admittedly, the Core i5-6600K isn't much of a test at stock speed, so I increased the load by overclocking the processor by 400 MHz, and choosing to sync the cores to force this 4.30 GHz speed on all four threads in the next round of benchmarks.
Here the results scaled nearly identically, with the Ninja 4 actually beating the NH-D14 pretty easily to widen the gap at third place overall in this group. Here the medium and low fan speeds offered the expected difference in cooling performance as well. You might also notice that even at the lowest fan setting, the Ninja 4 was nearly 9 degrees cooler than the baseline DEEPCOOL Gabriel low-profile air cooler, which finished last.
Noise Levels
The Ninja 4 was too quiet to register above the 33.4 dBA noise floor at low or medium
Here the Ninja 4 shows is just what it was designed for. This is as silent a cooler as I have ever installed or tested, period. It literally didn't sound like it was on at the low fan setting, unless you put your ear almost on the fan. Virtually no change was audible at medium, and high was just barely louder. And the numbers you see in the chart above don't do it justice; the 33.4 dBA number is the ambient level in the room.
Without better equipment and a quieter environment (such as an anechoic chamber) I can't tell you exactly how much noise the Ninja 4 produces, but I can say you will not hear it running in your system. It was somewhere below 33.4 dBA, and to my ears inaudible from the 24-inch distance I position the sound pressure meter. Incredibly impressed here.
Conclusion
There isn't much left to say after seeing the noise results. It is the quietest cooler I have ever tested, and performs very well with CPU thermals. The Ninja 4 was designed to be a silent cooler and it accomplishes this with flying colors. The mounting system works well, installation was straightforward, instructions were clear. Given the performance I literally can't fault a single thing with this cooler. Outstanding!
Thanks so much for the
Thanks so much for the review. Glad Maury is not the only one there who appreciates large air coolers and does not feel the need to ridicule the multitude of enthusiasts that have used them for years without broken boards from the weight.
I am shocked that a 0.3 decibel level difference between the high and low fan setting, negligible, results in a 10 degree improvement in “stress” temps. Hard to believe that is possible, but there it is. Why would one ever bother to run it at less than high with that kind of gain for almost nothing?
The results were a little
The results were a little skewed based on noise floor. I haven't had anything quiet enough to make my 33.4 dB ambient the limiting factor before this, and the low was somewhere below this. In reality there was more than a 0.3 decibel difference, but noise was really minimal at the highest setting.
Inside any reasonably quiet case you just aren't going to hear this, even on high, so I'm with you there.
I’ve had low speed fans
I’ve had low speed fans increase noise drastically by inducing resonance into the case.
Having said that, it’s probably rare and these noise figures make me think maybe a SINGLE SPEED fan would be worth considering because aren’t they possible to make cheaper, or at least last longer?
I wonder how it compares to AMD’s new single-speed fan as that thing baffles me. It has one speed but it’s audible.
On my NH-D14 I took off the 12cm fan and attached it to the rear of my PC. It didn’t add much to the cooling versus being off but replacing my cheaper case fan made a huge noise difference.
I have another recent Scythe
I have another recent Scythe heatsink with basically the same fan (the Scythe Kotetsu, with a 120 mm Glidestream PWM fan with the same rpm range, just without the switch), and let me tell you, there is a big difference in noise between ~800 rpm and ~1500 rpm for the fan, even closed in a case. Also a very noticeable one between ~800 rpm and ~1150 rpm. Even the ~800 rpm speed would be audible in my build at idle when things are quiet around the house; fortunately, the PWM control can keep it around 500 rpm then.
I will be installing this
I will be installing this cooler in my system soon. To be ready for the summer, I have an air conditioner but will not be on all the time during the summer. We will see how the Ninja 4 performs.