Testbed and Methods
Test Setup

• Case: Vigor Force
• Power Supply: Cooler Master Real Power Pro 850
• Motherboard: Foxconn N68S7AA
• Processor: Intel Core 2 Extreme QX6700, varying MHz
• Hard Drive: Western Digital WD2500KS 7,200RPM
• Video: NVIDIA GeForce 8800GTS
• Sound: Creative X-Fi XtremeMusic
• Physics: ASUS PhysX P1
• Memory: 2048MB (2x1024MB) G.Skill PC2 6400
• Optical Drive: Lite-ON SHW160P6S05
• Cooling: Vigor Monsoon II, OCZ Vindicator, Intel HSF (stock), Cooler Master Sphere

Hardware Settings

• CPU Voltage (before droop): 1.35V (stock speed), 1.625V (overclocked)

Software Configuration

• Operating System: Windows XP Professional with Service Pack 2
• Video Driver: NVIDIA ForceWare Version 158.22 (May 17 release)

Our current testbed for coolers is based on NVIDIA’s 680i platform that is known to have an excellent capacity for overclocking quad core Intel processors. The nTune utility that is available as an add-on to the nForce driver suite is also a fairly good monitoring tool, though we would not say the same for its other features. Unfortunately, all of the above holds true for motherboards based on NVIDIA’s reference design – those from eVGA, XFX, and BFG. Unfortunately Foxconn, the same company that OEM’s the NVIDIA reference design (and subsequently the motherboard for each of the aforementioned companies), has chosen to depart from this reference design with their N68S7AA. The main change lies not in the physical configuration of the board, though that is slightly augmented for the better, but the BIOS. Foxconn’s use of the American Megatrends BIOS is definitely a step in the wrong direction, as overclocking options (thanks in large part to Foxconn’s insistence on including the FOX Central Control Unit) are severely crippled on this version of the board. The motherboard is not “bad” by any stretch of the imagination, it is just not ideal for our purposes in this testbed. FPSLabs usually just plays with the hand it’s dealt, and that’s exactly what we’re doing here. When we get a chance to move to a more standard motherboard for the cooling testbed, we will certainly jump all over it. Furthermore, until we get some thermocouple thermometers to get truly accurate physical temperatures at repeatable locations on the CPU in the general area, we will be using software monitoring. To better provide for software testing, we are going with Intel’s Core 2 Extreme QX6700 processor, as the Core 2 family has a fairly reliable and accurate on-die temperature diode for each core. The program we will be using to monitor the temperature in Windows is RightMark CPU Clock Utility. This utility not only measures load and temperature on each core independently, but also provides a graph to record temperature variance over time.

To generate the load, we will be using an application called BURNK6. BURNK6 was developed in the days of AMD’s K6 core, but works conveniently well with Intel’s latest processors as well. BURNK6 has been proven by industry professionals such as Gabe Rouchon of Swiftech to be extremely effective at loading the CPU in a manner from which the greatest amount of heat is generated. The load generated by BURNK6, which is available as part of the CPUBURN package, remains constant over time, rather than slightly fluctuating such as with Prime95 and/or ORTHOS. A separate instance of BURNK6 will be used for each core on the processor, in this case four. Temperatures will be read from the RightMark CPU Clock Utility after 10 minutes of load (and idle) and the average temperature across all four cores will be reported. Temperatures will be recorded at idle and 100% load for each cooler tested in a given review. The processor will be overclocked to the point of failure with each cooler, and the maximum clock speed will be recorded.