Thermal Performance

Our temperature testing was performed on two different configurations. For the first we set the CPU to its default specifications of 3.33GHz and disabled Turbo Mode as well as power saving. The cooler used for this test was the Akasa Nero, a mainstream cooler which uses a 12cm 1600rpm fan which allows us to demonstrate the performance for the average user under normal conditions. The fan was manually set to full speed for the duration of our tests.

For our second test we overclocked the CPU by 1GHz to 4.35GHz (29x150mhz) using 1.45v and attached CoolIT's Domino cooler, a configuration which should give a fair indication of how the different pastes perform in an enthusiast system. The Domino was set to Maximum Performance.

For the idle test the system was allowed to sit at the desktop for 30 minutes with no activity. Load temperatures were recorded during sustained rendering with Cinebench R10.

Looking at the 3.33GHz results first we see that when the CPU is idle Arctic Cooling’s MX-3 improves our CPU temperature by 2°C when compared to the older MX-2. When the CPU is under full load the temperature increases by 20°C on both pastes which once again gives the MX-3 the edge. It should be noted that although the MX-3 hit 64°C, it did so very briefly and for the majority of the test it sat at 63°C, a 3°C improvement over the MX-2.

Throughout the 3.33GHz test the GC-2 outperformed the MX-2 and was never far behind the MX-3. Gelid’s latest paste also gave some excellent performance when at 4.35GHz under load where it once again outperformed the MX-2, this time by two degrees Celsius. Unfortunately the shine is taken off this performance a little by the fact that the MX-3 manages to outperform the GC-2 by a single degree at extreme speeds with the CPU fully utilised.