All tests were conducted at an ambient temperature of 20.6° Celsius, plus or minus 0.1°.
All three systems provided the same CPU temperature at low fan speed. The Koolance and Zalman pre-configured systems dropped CPU temperatures by one degree at maximum speed, while our own assembly was only designed to run at a single speed.
Because our own assembly was not designed to provide airflow over the memory or chipset, we chose the Antec Spotcool fan from the beginning. This gave our design a significant lead in chipset temperatures, but the same fan could also be added to either of the competing products. The real winner here is Zalman, because the huge side fan of its LQ1000 was capable of cooling the chipset to 39° in near-silence, without requiring additional components.
The primary graphics card ran the hottest in our SBM chassis, since both the Zalman LQ1000 and Koolance PC5-1326SL are designed for better airflow. Koolance has the overall best cooling performance, but at significantly higher noise levels.
The Zalman LQ1000 kept the lower card cooler, which seems strange since the Koolance PC5-1326SL blows its radiator fans directly into the graphics card, and here’s the catch : both graphics temperature charts only show low fan-speed results.
At full fan speed, the Koolance PC5-1326SL dropped temperatures of the second graphics card by over 20°, but provided only around 2° additional cooling to the primary card. The Zalman LQ1000 dropped the temperatures of both cards by around 3°. Any card between the graphics card and radiator will block the down draft of the Koolance radiator fans, so it’s best to simply keep in mind that a temperature drop of somewhat more than 20° will benefit only the uppermost card at maximum fan speed.