Testbed and Methods

Zalman VF2000 LED was tested in identical conditions on the graphics card and CPU. We assembled two test platforms for our performance tests:

Operating system, graphics card driver and other software were identical for both platforms (except for the chipset driver): Windows Vista Ultimate Edition x86 SP1, AMD Catalyst 9.6 and DirectX End-User Runtimes (March 2009).

First I’d like to say a few words about the testing methodology on the VGA card. The graphics processors and the graphics card as a whole were loaded in two ways. First we used 15 runs of Firefly Forest test from the synthetic 3DMark 2006 graphics suite in 1920x1200 resolution with 16x anisotropic filtering. This test mode will allow us to determine approximately what temperature mode will the graphics card experience in games. In the second case, we used FurMark utility version 1.6.5 in stability test mode to load the graphics card in 1280x1024 resolution. The tests here were performed for about 15 minutes up until the temperatures of the graphics processor and voltage regulator components became fully stable at their maximum point. We used RivaTuner v2.24 (by A. Nikolaichuk aka Unwinder) to monitor frequencies and temperatures.

Now a few words about the testing methodology on the CPU. We heated up AMD processor with OCCT Perestroika version 3.1.0. We ran a CPU test for 23 minutes, when the first and fourth minutes the system was idle to ensure that the temperature had time to stabilize. We monitored the CPU core temperature using CoreTemp utility of the latest version available at the time of tests. The stabilization period for the CPU temperature between the two consecutive test cycles was about 10 minutes. We took the maximum temperature of the hottest processor core of the four for the results charts. The ambient temperature was checked next to the system case with an electronic thermometer with 0.1 °C precision that allows monitoring the temperature changes over the past 6 hours. During our test session room temperature was unusually high and stayed at 26.5-27 °C.

During this test session we managed to overclock our AMD Phenom II X4 940 Black Edition processor used for Zalman VF2000 LED tests to 3.52 GHz by changing the unlocked multiplier. The nominal processor Vcore was increased to 1.45 V in the mainboard BIOS.

This is very modest overclocking, just as the cooler itself.

As for the Zalman competitor, things are not that simple either. During the graphics card tests we can easily compare it against a reference Radeon HD 4870 cooling solution with not only high level of generated noise but also high cooling efficiency. However, when it came to CPU tests it turned out not that easy to find a proper rival for Zalman VF2000 LED. I didn’t have any low-profile hybrid coolers at my disposal that is why it didn’t make sense to perform the comparison from this standpoint. The only criterion left was the price, and here we could consider Scythe Kabuto retailing for $48 to be a possible opponent.

So, we will use it for our comparison today. Of course, Scythe Kabuto can’t be installed onto a graphics card and it is definitely not a low-profile cooler, but its price is comparable to that of Zalman VF2000 LED. Besides, both these coolers create similarly directed airflow.