How We Tested
| CPU Cooling - Overclock and Normal Performance Settings | |
|---|---|
| Processor | AMD FX-8350 @ stock settings (normal) |
| AMD FX-8350 @ 4.4GHz, 1.4325V, LLC=Medium (overclocked performance) | |
| Motherboard | Gigabyte 990FXA-UD7 (BIOS F10) |
| RAM | 1x 4GB G.Skill (DDR3-1333 CL9) |
| Graphics Card | XFX Radeon HD 5450 (passive) |
| SSD | 60GB Kingston SSDNow V+ 200 |
| Power Supply | Xilence XQ Series Platinum R4 (1000W, semi-passive) |
| PC Case | Cooler Master CM Storm Stryker |
| Case Fan Front and Back: 1000 RPM | |
| Case Fan Top: 900 RPM | |
| Other | Aqua Computer Aquaero 5 LT (firmware 1027) |
| Arctic MX-4 (thermal paste) | |
| Operating System | Windows 8 Pro 64 Bit (Version: May 2013) |
In order to measure the processor’s temperature, it first endures 60 minutes of Prime95 with custom settings, heating it up. Then, its core temperatures are measured for 10 minutes. Finally, all of the temperature measurements across the entire interval are averaged.
This procedure yields more accurate results than recording the temperature at just one point in time. Unfortunately, the processor’s internal sensors provide relatively imprecise values, since they exclusively use whole numbers. This means that the results might have a somewhat higher measurement error.
Noise measurements are conducted with an open PC case from a distance of 30cm perpendicular to the center of the CPU cooler. Don’t panic after looking at the results; a closed case and a more reasonable distance from it will significantly reduce the coolers’ noise level. A greater distance would have meant that ambient noise would play a larger role, increasing measurement error.
We’ll adjust our benchmark system for ITX-oriented CPU coolers soon so that we can test them in a fitting PC case. Even without a purpose-built platform, Raijintek's Pallas turned out to be a strong contender in our tests. It didn’t just manage to complete our usual 125W CPU suite without breaking a sweat, but it also left us amazed after seeing the results of our overclocked 180W workload.
It’s not every day that a diminutive 6.8cm CPU cooler, which was designed for cramped spaces in tiny PC cases, can take on established tower coolers priced at up to $50. We didn’t even trust our results at first, but several more test runs didn’t change them.
The Pallas’ main competitor proved easy to find: it’s the Thermalright AXP-200, which is our go-to cooler for extremely high-performance systems that are supposed to be built with a very small form factor. Alternatively, we use the Deepcool Lucifer, which operates on a similar level as the HR-02 Macho, and the be quiet! Shadow Rock 2, which might not be able to provide quite the same cooling performance but has a leg up on the competition when it comes to noise level.
With the exception of the Thermalright AXP-200, we’re putting Raijintek's Pallas up against supposedly stronger tower coolers in its price range. Again, the results turned out to be a big surprise!
