Before the end of this roundup, let’s summarize comparative graphs, with results for every cooler. Charts were ranged in three categories: GPU, Ambient and MOSFET. Here are detailed summarized results of fights of all five competitors.

First there are charts sorted by GPU temperatures in idle state for not oveclocked and overclocked graphic card.

Every cooler is in graph represented by one color. Differences in quality are remarkable, last two tested coolers are clearly leaders in this group, and active coolers have no chance to compete here. Thermalright and Accelero go throught the test side by side. Difference appears with installed 92 mm fan, here is Accelero gaining its position. Thermalright with recommended 92 mm fan set running at moderate speed have difficulties to compete wit S1. Accelero S1 belongs apparently to a different category.


Here are graphs with GPU temperatures in high load – with test GT2 – Firefly Forest, from 3DMark06 application. Again measured in both states – not overclocked and overclocked graphic card.

These graphs are probably most interesting. There are no changes in trend, in accommodation to “idle state graphs”. Colors distribution is similar; there is a positive change for Zerotherm GX815, which has completely lost in previous round. This time, with not oveclocked graphic card GX815 is closely behind Zalman, and after overclocking it reaches Accelero S1 level. But Zalman’s cooler took a new breath before the end, and Thermalright’s cooler was loosing again. HR-03 needs to be supported with recommended 92 mm fan set running at full tilt to compete with S1 working in passive mode. I guess this makes quite a difference. There is a huge difference of 32°C between S1 and HR-03 working in passive mode after overclocking, which is a very poor result for HR-03. Accelero S1 did incredibly well this time, as though without direct support of additional fan left all its competitors clearly behind. Active coolers do not find help in their fan, set running at maximal speed and with a big roar, they are futilely trying to reach super-silent S1.
As it was proved with measurements, Accelero is showing the same great performance, with two fans of both sizes installed, no matter if 92 or120 mm. There is notable difference, with one fan installed. 92 mm fan is able to compete with 120 mm fan-regulated at minimal 800 rpm, only when it is set running at maximal speed. The situation is similar with both fans installed. Regarding 120 mm fan, there were no differences in cooling efficiency, whether one or two fans were fixed. Divergence appeared, when 92 mm fans were tested. There seems to be quite a difference, comparing set up with one 92 mm fan and two 92 mm fans.
HR-03 did not disappointed in competition with coolers with preinstalled fan. However, its overall performance was rather frustrating. I would say, that this cooler is usable only with 120 mm fan, even its pre-set is surprisingly not supported by manufacturer. That is weird, as how I have already said; performance of HR-03 with 92 mm fan is not any special.


Following charts are sorted by ambient temperature, tested in idle state and later in high-load state.

I will leave graphs without comments, as looking at charts, you can easily follow all important information and like that you can make your own opinion about presented coolers. What is more, ambient temperature data are not significant for choice of a new cooler.


To conclude, following charts are sorted by MOSFET temperature. Again, data were recorded in idle-load state and in high-load state. First graph presents system in idle-load state.

In the previous chapter, heat-accumulation in MOSFET area, related to Accelero S1 was already mentioned. Here it is well documented. Accelero S1 – in passive mode, is the last one in our benchmark. However, with fan support, its performance is much better, or I should say, with support of two fans. Zerotherm GX815, with its radially shaped fan, not able to make any serious draught around MOSFET, is dragging behind as well.


…at the end we have charts sorted by MOSFET temperature in high-load.

To summarize, passive coolers are clearly taking last positions in this benchmark, well, in this contest, only passive equipment is rather unsatisfactory. On the other hand, with fan set running full tilt, situation completely changes. Doped with a solid draugt, they are fully capable to cool MOSFET, and there is no definite winner at the end. I have to add, that behaving of HR-03 with 92 mm fan was rather suspicious. Even GX815-with its not very smartly positioned fan, was after MOSFET overclocking, showing better results, while voltage regulators using HR-03 with 92 mm fan were “baked”. Luckily with 120 mm fan support HR-03 does much better. VF900 did not get lost at all. And this time, also stock cooler profits of its smartly designed aspiration, and shows its qualities, unfortunately well supported with typical noise-reminding exhaust-pipe.