Temperature, Noise, And Acoustic Efficiency
We tested the thermal properties of each case at maximum CPU and near-maximum GPU loads, using 11 threads of Prime95 and looping 3DMark 11's first test simultaneously. We locked the fans of our graphics cards at maximum speed to gauge the peak cooling performance without any added variability.
A well-placed intake fan and wide open side panel allow Aerocool’s Strike-X ST to lead at its highest fan speed. Azza takes second, place while NZXT follows close behind. Note that the Hurrican 2000’s fan controls are hard-wired to its top panel fans, which were replaced as part of our cooler installation.
We continued to use the fastest GPU fan speed through our full-load noise-level tests, but let those fans spin down completely for our idle tests. Mesh side panels usually fare poorly in this test at full GPU fan speed.
Remembering that low-speed fan mode gave Aerocool’s Strike-X ST a last-place finish in our temperature charts, we weren’t surprised to see a low noise level at idle. The noise under a full load is a completely different matter, as the same unit’s mesh side panel allows nearly 100% of the GPU cooler noise to pass through.
NZXT edges out Azza, and users who want the best of both cooling and noise can even try connecting their fans to thermally-controlled motherboard headers.
A comparison of temperature to noise should be a matter of simple division, except that we want the case with the lowest temperature to have the highest score. To achieve this inverse scale, we first divide the average temperature of all cases/all tests (52° C over ambient) by the average temperature of each individual case. We then divide the average noise level of each case by the average noise level of all cases (43 dB). Finally, we divide temperature (inverse scale) by noise (direct scale).
The resulting relative efficiency chart relates each case’s performance to the average of all tested cases. For example, at 105.4% of average, Aerocool’s full-fan-speed Strike-X ST performs 5.4% better than average.
