Four different presented high-end coolers are tested in our roundup (than there is stock cooler for comparison). Two of the coolers: Zalman VF900-Cu and Zerotherm GX815 are active coolers, with installed fans, so the test setup was adapted. Their performance was measured at four different fan speed rotations and input supply volatge was set up with manual regulator on values 5, 7, 9 and 12 Volt. Surprisingly, values of fan speed rotations ere nearly identical for both of coolers, deviations were insignificant. Just for curiosity, coolers were tested also in passive mode, with fans dismounted, keeping in mind that they are not designed for this purpose.
Other two coolers: Accelero-S1 and Thermalright HR-03 are primarily made for passive cooling. They were tested according to this, but anyway how it was already mentioned, we measured their performance also with support of two different fans, 92 mm and 120 mm. We even tested the S unit in two ways, with one or two fans installed. For testing, high quality Coolink fans were used: 120 mm model SwiF 1202 and 92 mm model SwiF 921. Both models have similar fan speed rotation; we used them in mode 800, 1200 and 1500 rpm. As you can see, to make a complex idea about VGA coolers, loads of measurements were made.

Fundamental in this roundup is graphic card ATI Sapphire X1800GTO2, described in previous chapter. Graphic core R520, based on 90 nm technology, served as a heat supplier.
Function of graphic card at basic clocks is following: core – 500MHz at a voltage 1.075 V; and memories 500MHz at a voltage 2.097 V. All tests were setup at these values. Overclocked graphic card: core 650 MHz with increased volatge 1.40 V and memories 700 MHz with voltage 2.097 V. As you can see, test conditions were rather harsh.

Our test rig is described above. There were two regulated supporting 120 mm coolers used, to strengthen the air fluctuation inside the case. SilentBlade - forced draft blower, showed after the speed regulation 750 rpm, at 5V. Noctua NF-S12 1200 fan was installed on a processor cooler. Its speed rotation was software set up, at fixed 900 rpm. For stalling the graphic card (first of all), programme 3Dmark06 was used. Its set up was based on ten-times repetition of graphic test GT2 – Firefly Forest. After long-term observations, I have realised that exactly this test is able to highly stall the graphic card.

System of testing was following: After starting a computer, I left the computer’s temperature to stabilize in stand mode. Than the system was tasked with already mentioned programme. After its finishing I left all components and graphic card to cool down for 15 minutes, until the temperature was stabilized again. This action was repeated for all measurements in a same way.

Temperatures of graphic card were registered using a programme RivaTuner, concretely its function „Hardware Monitoring“. This programme is able to register also Core temperature, Ambient temperature and Voltage regulator temperature. These three values were than evaluated and analysed.