Part 2. Test results and summary

We use the same procedures for evaluating thermal efficiency of coolers for Socket 754/939/AM2 as for Socket 478 coolers and those that are actively used for testing LGA775 coolers. The primary data used for the consequent calculation of thermal resistance are temperature readings of the thermal diode built into a processor. What differs is the thermal source (Athlon 64 3700+ processor), the base platform (Fujitsu-Siemens D1607 motherboard), and a set of test applications.

Testbed configuration:

• Fujitsu Siemens Computers D1607-G motherboard
• AMD Athlon 64 3700+ CPU (Clawhammer)
• OS: Microsoft Windows XP

We use the S&M utility to simulate maximum thermal load of a processor, and SystemGuard from Fujitsu-Siemens to monitor temperatures.

Chart 1. CPU Core Temperature

Notes
Each cooler is tested with Stars 420 thermal grease
The diagram contains complex results

Chart 2. Thermal resistance

Note
Thermal resistance θ_ja is defined as the relation
θ_ja = (T_j - T_a)/P_h, where T_j is the temperature of a CPU core, T_a is the environment temperature (it's 33°C in this case), P_h is the thermal capacity of a processor (in this case it's 80 W).

Finally, at the end of this article we publish the noise measurement results (the test method is described in the article Cooler noise characteristics and the method of measurement) as well as the efficiency/noise rating of coolers.

Chart 3. Noise characteristics, A-weighted sound level

Note: Background noise level 18 dBA

Chart 4. Efficiency/noise rating

Note
The efficiency/noise ratio (ENR) is calculated as:

ENR = DM*(R_t/TC)/(NL/R_n), where

R_t - reference temperature (the reference thermal resistance θ_ja of the cooling system - 0.3°C/W), TC - the core temperature with the operating cooling system, R_n - reference noise (the reference noise level is 25 dBA), NL - noise level, generated by the cooling system, DM - denominate multiplier 10).

Table 1. Coolers statistics summary