To thermal test the coolers we mounted them into our new Core i7 test machine, equipped with an Intel Core i7 Extreme 965 overclocked to 3.6GHz – the absolute limit of the stock cooler’s abilities.
When mounting the coolers we used our own quick cure Lumière thermal compound from Arctic Silver rather than the TIM supplied with individual coolers to create a level playing field with which to judge performance, using the cling-film technique we recently documented to evenly spread a thin layer of TIM evenly over the CPU’s integrated heat spreader.
With the machine booted we then used the small FFT torture test in Prime95 to fully load all eight of the CPU’s threads, producing the maximum 130W thermal output of the CPU and used Core Temp to monitor the individual temperatures of all four cores.
After thirty minutes of torture testing, we then recorded the average core temperature of the CPU and, by comparing this value to that of the ambient room temperature throughout the test, we achieve the Delta T values which you see below.
We decided to test some the coolers multiple times under various circumstances, such as removing the shroud on the Zalman CNPS 9900 to see if Zalman’s claims are true or removing the second NF-P12 from the Noctua NH-U12P to show current owners what to expect if they upgrade.
We’ve also included the results for the Intel Reference HSF and the mighty Thermalright Ultra eXtreme 120
, the recipient of our Best Cooling Product 2008 Award
, as the LGA 1366 conversion kit for it can now be purchased for around £9. Can any of this new crop of super coolers dethrone the TRUE for brute force cooling, or is its six heatpipe design still the best after over two years on the market?
Thermalright Ultra eXtreme 120
Cooler Master V8 (Full Speed)
Noctua NH-U12P (Single Fan)
Noctua NH-U12P (Dual Fan)
Cooler Master V8 (Half Speed)
Zalman CNPS 9900 (With Shroud)
Asus Triton 81
Zalman CNPS 9900 (Without Shroud)
Intel Stock HSF