To test CPU Coolers we simply boot the PC into Windows XP and measure temperatures under idle and load states. The onboard temperature sensor is disregarded and instead the CPU’s own diode is used.
For idle testing, the rig is left doing nothing at all for approximately half an hour and then we take the most representative reading at the end of this time period. For load testing, the same procedure is used except both cores are loaded to 100% using four processes of the software CPU burn-in (as the Q9450 has 4 cores that each need loading).
Ambient temperature was 20°C. For comparison purposes I will be comparing the cooler to another Scythe product – the Scythe Kama Angle – and the Stock Intel Cooler. The cooling inside my NOX Coolbay HX case consists of 2 x 200mm fans at low fan speeds alongside a 120mm rear mounted Noiseblocker Multiframe M12-S1 fan @ 550RPM.
I will also be testing the cooler using a number of thermal pastes:
- Gelid GC1
- Noctua NT-H1
- Coolink Chillaramic
- Scythe Scyte-1000
As well as setting the included fan to different fan speeds to see how it performs. To do this I will be using the Noctua LNA and ULNA adapters do restrict the voltage so as to reduce fan speeds.
Finally, I will be using different fan setups to look at how this affects performance.
All of the tests will be conducted using the Scyte-1000 Thermal Elixer except of course the thermal paste comparison.
First up I will be testing the Mugen 2 to the Scythe Kama Angle and Stock Intel Coolers. To do this I am simply plugging in the PWM connector straight to the motherboard.
|Scythe Mugen 2
|Scythe Kama Angle
|Stock Intel Cooler
Using Speedfan I could see exactly at what speed the Mugen 2 was spinning: at idle it was roughly 400RPM going up to 650RPM at load. This seems strange as surely you would expect the fan to be spinning a lot faster at load considering it can go up to 1300RPM. Perhaps this is because Scythe’s PWM graph is linear and so only reaches maximum speeds at say 50°C, although I cannot verify what their PWM graph looks like as I couldn’t find any information. It would have been nice to see a much more rapid increase in fan speed as the CPU reached 100% load so that the cooling was even more impressive.
As it stands though, the Mugen 2 manages to beat the Kama Angle by a couple of degrees and by a much greater margin, the Stock Intel cooler (up to 10 degrees at load).
Next I’ll be looking at fan speeds and as I said before I will be using the Noctua LNA and ULNA adapters along with a molex connector in order to set the fan speeds at 500, 750 and 1300 (max) RPMs.
|Molex Adapter @ 1300RPM
|Noctua LNA Adapter @ 750RPM
|Noctua ULNA Adapter @ 500RPM
Evidently whacking the fan speed up to max has the desired effect with the Mugen reaching very impressive temperatures in idle and load states at 30 and 41 degrees respectively. Again, as expected there are a couple of degrees drop down at 750RPM.
However, the surprising thing about 500RPM results is that they are virtually the same as the 750RPM set with the only difference being one degree at idle. This seems odd so I retested just to make sure but again saw the same thing. Overall, very impressive temperatures throughout though and either of the settings would be adequate for most users.
To test out what is the best fan setup is and whether the Scythe fan lives up to its manufacturers claim of being the best fan for the job, I chose to test a few different setups:
- Single Scythe fan
- Single Noctua NF-P12 fan
- Dual Scythe fans at either end
To ensure the test was fair I used a molex connection for each fan thus running them at maximum speeds (max for all fans = 1300RPM).
|1 x Scythe fan
|1 x Noctua NF-P12
|2 x Scythe fans
As you can see, whichever fan used on its own did not affect the temperatures, so in fairness Scythe probably are correct in saying their fan is the best to use, although other 120 fans can produce equal readings.
However, and rather disappointingly, the dual fan setup did not yield as impressive results as I was hoping for. In fact it only shaving one degree of idle and load so for the added noise outputs which with both fans running at 1300RPM is rather considerable, I wouldn’t recommend this setup.
Across the board, the temperatures did not vary too greatly with only a couple of degrees in it. However, if you’re looking for top performance then the Noctua NT-H1 and Scyte-1000 came top. This is impressive from Scythe as the NT-H1 compound is a renowned performer and so to keep pace with it, is an achievement most certainly.
That concludes the testing and it is evident to me that for the majority the Scythe Mugen 2 has excelled. Overall, if looking for the best performance I would chose to use the Mugen 2 with a Noctua NF-P12 fan and the Scyte-1000 Thermal Elixer as this seemed the best combination with it not being really worthwhile using dual fans due to the added noise outputs.
However, the NF-P12 does bring an added cost so if you’re looking for great performance at a lower cost, you may as well just stick with all the included parts and they seem to do a pretty darn good job.
Using the PWM connection, I didn’t really notice any extra noise but then again at load the fan only got up to 650RPM so it wasn’t too likely that I would. However at 1300RPM the fan is really quite noisy and even more so with dual fans.
The best fan setup in terms of noise was using the NF-P12 fan and so I would recommend using this fan it’s not only less noisy but provided identical results to the Scythe fan.
QuietPC list this cooler at £43 inc. VAT which puts it near the top end of cooler pricing. It’s well worth it for the cooling ability, but you’ll have to dig a little deeper than you might like to to purchase it.