Hello guys and girls.
To start with, I've always been obsessed with fans, trying to get the best ones in my builds. I have looked around and did not find any websites that dedicate themselves to test just fans and rightly so, It wouldn't be that lucrative.
I am in the process of building a website dedicated to just that, maybe expand it later, but fan testing will be the top priority. For now though, I'll stick to OCN's forums.
I have begun testing and I will be updating this thread regularly with the data gathered from these tests.
I plan to build several different rigs to test different sized fans, the one I'm starting with is dedicated to 120 mm fans. The heatsink used in each system will remain the same. The only thing that will change are the fans.
As stated earlier, I only have one spare system that I've dedicated to test 120 mm fans. Hopefully, with enough funding and support, I will build at least two more testing rigs, one for 140 mm fans and another for 92/80 mm fans.
Righto, down to business then, let me walk you through the 120 mm testing rig:
Processor: Intel Core i7 920 (C0/C1 Revision)
Motherboard: Asus Maximus II Extreme
Memory: 3 x 2GB (6 GB) Mushkin Redline DDR-3 1600 MHz
Graphics Card: AMD HD5450 (Passively cooled)
SSD: Corsair Force 3 60GB
Power Supply: Seasonic X-Series SS-460FL (Fanless)
Chassis: Lian-Li Pitstop PC-T60
CPU Cooler: Prolimatech Megahalems Rev.C Black Edition
Thermal Paste: Xigmatek Freezing Point G4718
Prolimatech Sent me a Megahalems Rev.C Black Edition as well as a set of their Red Vortex 12 LED fans, rated for 1600 RPM. On top of those, they've also sent me a set of extended mounting clip to use with thicker 38mm fans and a set of screws required to use this cooler with a LGA2011 motherboard.
Those are the meat and potates of this setup, now let me reveal the BIOS and windows settings:
First of all, the overclock. This CPU has sort of been a pain in the bum as it's overclocking capacity is severely limited. It shouldn't matter though as what's more important is the amount of voltage I'm feeding it. It's running at 3.6 GHz using 1.4-1.42 volts (according to CPU-Z). In this trim, using Prolimatech's fans in push/pull, and with an ambient temperature of around 26 C, the CPU is running at an average of 78 C. Not bad at all as I was expecting somewhere around the 85-90 C mark. The Base Clock / Multiplier ratio for CPU is 180 x 20.
The memory is running at 1442 MHz. The timing for this set of memory is 6-8-6-24 with 1.65 volts, but I didn't manage to get it stable with those timings. As a result of that, I'm using 7-8-7-24.
In the BIOS, I've pretty much disabled anything that is not necessary for the overclock such as Hyper Threading, Virtualization, etc (They are sort of relevant as you might need more voltage to run these features). I have gone one step further and even disabled LAN (Ethernet port) and some other irrelevant features.
The motherboard is using the latest BIOS revision (2101). All spread spectrum settings are disabled. The Load line Calibration is turned on. The voltage set in the BIOS is 1.3875, but as stated earlier, it goes up to 1.42 volts in windows.
The operating system is Windows 7 Ultimate 64-bit with the latest updates installed (as of june 25th, 2012).
The torture testing will be done with Prime95 64-bit version 27.7. The FFT size will be 1280K (min and max) to keep the amount of heat generated the same everytime. 1280K was optimal since it used the most power during full load. The test is ran for 5 minutes every single time.
CoreTemp 0.99.8 will be used to monitor the temperatures in each session. An average will be recorded after each test and the ambient temperature will be deducted from that value.
I will try to be as accurate as possible to get the most consistent results. The heatsink and the whole system will be cleaned with a DataVac electric duster before and after each test.
Before running the main torture test, Prime95 will run for 5 minutes to warm the heatsink, after which the main test will commence.
As most of you would know, the performance of any certain fan, depends entirely on where and how you use it. Different heatsinks/radiators have different thickness and fin densities and it's near impossible to test a set of fans in every possible scenario. In my series of tests with 120 mm fans, Prolimatech Megahalems will be used. The distance between each fin layer is 2 mm which is pretty much the same as most other CPU Coolers on the market. I've measured the fin gap on a Noctua NH-C14 and a Cooler Master Hyper 212+ and they are exactly the same as the Megahalems.
Now then, everything I just wrote pretty much involved temperatures and not much else. Worry not though as I will also be measuring the noise levels of each and every fan and maybe later, the voltage required to run these fans at full speed. The chart that I'm going to make will include a simple yet intuitive Performance to Noise ratio section alongside other standard bits and bobs.
I have a variety of tools (Multi-meters, Decibel-meters, etc) available. I will test each of them and determine which one is the most accurate to be used in my tests.
Update #1: I've changed the power supply for the 120mm test rig (Corsair TX850) to a Seasonic X-Series SS-460FL (Fanless) unit. The only moving part in the system will now be the fans on the CPU cooler, allowing accurate dBA measurements. (7/July/2012).
Update #2: I had to change the thermal interface to Xigmatek Freezing Point G4718 since Indigo Xtreme lost its consistency after exposure to high heat levels (CPU temperature over 90C). (28/Sep/2012).
Update #3: 140mm rig finished and first set of fans tested. (5/Oct/2012).
Update #4: GPU in the 120mm Testing rig was changed to a passively cooled HD5450 as the old one (GT 210) died. (20/Feb/2015)
Update #5: Moving forward, all 140mm fans will be tested on the 120mm test rig for direct comparison between the two sizes. (3/Mar/2015).