Testing Methodology
Testing Methodology
To ensure that the results from one review to another are not only reproducible but actually pertinent to this review, the Testing Methodology will be the same throughout all water cooling review. If something does change we will be sure to make a special note of it and explain why this change was done and more importantly why things had to be changed or altered.
Thermal Paste and Application Methods:
Arctic Cooling MX-2 thermal paste was used for all water based CPU cooling solutions during these tests unless otherwise noted. Application of thermal paste was in accordance with the TIM manufacturer’s instructions; and while not necessary, the TIM was allowed to cure for 24 hours under moderate to high loads (with periods of low loads) prior to testing.
Fans Used:
120mm:
For all water based CPU Cooling Solutions which do not come standard with a fan, a pair of Noctua NF-P12-1300s and a Scythe S-Flex “G” 1900RPM fan will be used if it accepts 120mm fans. With these two fans we are able to simulate different fan speed conditions as indicated below.
High Speed:
1900RPM Scythe S-Flex “G”. To be more precise our specific fan runs at 1860RPMs. Any stock fan which comes with the ability of being controlled by means other than the motherboard (e.g. remote fan speed controller, potentiometer, rheostat, etc) will be set to this speed during the High speed test and BOTH sets of performance results will be included.
Dual Fans:
Dual NF-P12-1300s
*Dual fans only used if the cooler comes with the necessary mounting hardware.
92mm Fan:
If the cooler being tested only accepts 92mm fans, a Noctua NF-B9-1600 will be used.
If the given CPU cooling solution comes with a stock fan we will also include its numbers in the closest of the main tests BUT we will also include our standard fan results in that particular tests.
Fan Notes:
- If a water cooling solution cannot mount an aftermarket fan, we will be only including the stock fan results. However, if the stock fan speed can be precisely controlled by means other than the motherboard BIOS (an included remote fan speed controller, potentiometer, rheostat, etc), the cooler will be tested at different fan speeds.
- For dual fan results ALL water coolers capable of mounting two fans (and come with the necessary hardware) will be tested with two NF-P12s and the Dual Fan graph will contain data for other such dual capable fan coolers.
We feel that the combination of multiple speeds and multiple fans will allow us to give you our readers clear and precise idea of the capabilities of a given unit, in an accurate comparison. It will also help eliminate the occasional “zinger” such as when a manufacturer includes an extremely high-speed fan in order to possibly offset poor thermal performance.
Case Temperature Testing:
To help give us a clear picture of what effects the water cooling unit in its default configuration will have on in-case temperatures we will be using a K-type A W Sperry STK-3016T single line Digital Thermometer with reading sensitivity of 0.1°C. It is positioned 1" from the top of the motherboard in the location indicated below:
<img src="http://images.hardwarecanucks.com/image/akg/water_cooling/H70/vent_sm.jpg" border="0" alt="" />
Unless specifically noted, the internal ambient case temperature testing will BEGIN with a “cold” system with the internal case temperature reading 20°C. While we base the CPU temperatures on 15 minutes of testing, the internal ambient air temperature test will consist of 45 minutes of running Prime95 “small fft” and at the end of the test the internal case temperature is recorded as the official result for that run. Each test run configuration was run at least 3 times and only best results are represented. If at any time during the 45 minute test the internal ambient case temperatures reached 45°C the test is considered a FAIL and recorded as such.
Environment:
Except where noted all comparison testing was done inside a closed case with a room ambient temperature of 24c. If at any time the room temperature increased or decreased by more than 1°C, testing was halted until the temperature constant was re-established.
Testbed:
The case chosen for this test bed is a Cooler Master 690. We chose the 690 as it a highly adaptable case with not only multiple door fan ports, but also multiple top fan ports as well as the typical single front and back fan ports. To populate these fan ports we chose Scythe E 120mm units with Sony Fluid Dynamic Bearings. Unless otherwise noted, only one of the top two exhaust fan ports, the rear exhaust port and front fan intake port will be populated. The rear exhaust port will be populated by the review item's fan and radiator where possible, for the air based cooling alternative used in the review as a counter example another Scythe E will be used.
Temperature Recording:
Recorded temps were as reported via the Real Temp plug-in for the RivaTuner monitor program.
Max and Average load temps are based on 15 minutes of running Prime95 “small fft” and are taken directly from RivaTuner’s built in capabilities.
The maximum temperatures will be the highest recorded temp displayed for any of the cores during the 15 minute test. While RivaTuner will display each core's average temperature it does not easily show the average of ALL the cores. To this end we will be simply taking the average of all the cores adding them together and then dividing by the number of cores.
If during any test temperatures of 95°C or more are displayed in RivaTuner (for any core) for more than 10 consecutive seconds the testing will be halted and that test run will be considered a "fail".
Idle temperatures are the lowest recorded temperature during idle period as recorded by the RealTemp Rivatuner monitoring program.
All CPU throttling technology was disabled in the BIOS; as was all CPU fan speed control. In addition, Turbo Mode was disabled and Hyperthreading was enabled. All power connectors for the review item are connected directly to Molex connectors to ensure they were running at full speed.
All tests are run a minimum of three times and only the best results are represented.
Charts & Graphs:
Each chart will include the Maximum or “peak” temperature we recorded, the average temperature and the idle temperature.
No passive results will be shown UNLESS manufacturer claims the ability to passively cool a processor. If a manufacturer claims passive capabilities we will include the performance numbers in the charts. The only exception to this is if the review is a “review roundup” and to keep the charts from becoming confusing we may not do so.
All water cooling reviews will also include a air based CPU cooling which best approximates the price range of the water cooling solution being reviewed. This way you will not only know how it compares to other water cooling units but also an Air based CPU cooling solution which is in the same approximate price range.
Sound Pressure Testing:
To give a more accurate and less of a personal opinion on the noise level of the stock fan which accompanies the heatsink, we have included a new section for sound pressure testing. These tests are done in our open case setup outlined above with the meter positioned 30 inches away from the cooler and mounted on a tripod. To ensure the background noise does not skew the results all tests will start by recording the ambient noise of the room. Only when it meets our standards will the testing commence.
To ensure that no external noise unduly skews the results, the GPU used will be a passively cooled unit and the only active fan will be the one on the cooler while the PSU and HDD are isolated away from the immediate area.
These tests are run late at night when no other people or animals are awake and thus unable to influence the results.
All fans are run at their maximum speed with no voltage or PWM control being used during the sound pressure tests.
The sound pressure meter used is a DT-805 which has been professionally calibrated and NIST certified. We will record the highest levels obtained with said meter and record it as our result. The test will be 15 minutes long and will be run while the fan is running full speed via a Molex connector and the CPU cores are under a full load via Prime 95 Small FFT.
Please note: The Scythe S-Flex G and Noctua NF-P12-1300 (at 1300 and 900rpms) numbers are taken when mounted to a Cooler Master Hyper 212+. We feel that it would be extremely unfair and unrealistic to include noise rating for these after market fans if they were NOT mounted onto a cooler. They are included to help give some sense of proportion to the charts and allow you to more easily compare a stock fan against a known quantity.
Complete Test System:
Case: Cooler Master 690
Processor: Intel i7 920
Motherboard: Gigabyte X58-UD3R
Memory: 6GB Aneon Xtune DDR3-1600
Graphics card: Asus 8800GT TOP
Hard Drive: 1x WD 320GB single platter
Power Supply: Topower Powerbird 900W
Special thanks to Direct Canada for their support and supplying the i7 920 CPU.
Special thanks to Gigabyte for their support and supplying the i7 motherboard.