Let’s test passive coolers now. Of course, not only passive mode was classified. Accelero S1, as the new cooler on the market, was tested first. I have to admit, that I had high expectations from Accelero S1. For its precise valuation, many measurements were made. This Arctic Cooling product was measured in passive mode, and with support of one and two fans, using fans in two different sizes – 92 and 120 mm. Measurement were made at three different rotational speeds: 800, 1200 and 1500 rpm. All around 26-test measurements were made, to get idea about this cooler, so as you can see we are not any cheap to make a judgement after one-two tests, like many others do.

Accelero S1 is more than serious contender while working in passive mode. In idle mode it is a leader of the roundup, it is knocking down even coolers with active support in high load, only stock cooler was slightly better with fan at 5000 rpm. With fan support, Acclero S1 is just a great performer. Results are similar for all set up rotational speeds, there are little differences, if measured with one or two additional fans.
In high load there is no change. Cooler’s passive performance is amazing. Accelero S1 performes great when graphic card is overclocked, and obviously it has absolutely no problems when the card is not overclocked. It is already clear; that S1 does not need fan support, however, also slightly better result is always better result and that is why I am for installation of additional fan support. As you can see in graphics, all fan combinations, apart from single 92 mm fan set up, make better result. There is no remarkable relation between rotational speed and recorded temperatures. Cooling effects with two 92 mm or two 120 mm fans, are identical in default load and after overclocking of the graphic card.

Recorded ambient temperatures have similar trend that measured GPU temperatures. Again fantastic results were obtained only with slightly higher rotational speed. The use of one 92 mm fan has probably no effect on better efficiency, as recorded results are nearly identical with values in passive mode.

I have found some serious problems in cooling of supplying regulators. Fins of cooler are placed closely above MOSFET passive and this is causing a heat accumulation in this area. Bigger dimension of aluminium thermal interface material, comparing to previous model, does not help. Here the need for istallation of additional fan is noticeable. Set up value of rotational speed is not crucial in this case; important is to make some draught in this area, which will cool down hot voltage regulators. Here with one cooler, capacity for cooling MOSFET down is obviously lower, and important is also size of a fan. F.e. using of 92 mm fan at 800 rpm, has no big effect.

The last high-end competitor in this roundup is Thermalright HR-03, nowadays considered as a leader in cooling solutions. It will need to make a great job, to compete with superb Accelero S1. Also in this case, measurements with additional fan support were made. Coolink - 92 and 120 mm fans were used as previously. The size of a heatsink gives limited options to install both fans in one. Rotational speed was regulated at 800, 1200 and 1500 rpm.

As presumed, HR-03 in passive mode has a difficulty to compete with S1. There is rather big difference in constructions of two coolers. Quite big differences appear already in idle state. HR-03 looses in idle state 5°C, after overclocking there is 11°C spread. After loading the system with 3Dmark, Thermalright HR-03 is able to cool only core of not overclocked graphic card. After overclocking, cooler is demonstrating difficulties and temperatures are critical. Installation of fan is a need.
HR-03 behaves identically with all combinations of fans installed. There is slight linear increase of temperatures in accomodation to rotational speed. In this point, Thermalright is able to compete with Accelero, and as by now a leader, it is knocking down also other coolers in this roundup. Also in high load it is doing well comparing to others, but not as well as S1 cooler. With recommended 92 mm fan it drags behind more than with 120 mm fan. When cooler is set running at low speed, temperatures of graphic core in overclocked state are rather higher than temperatures measured in idle state.

During measurments of ambient temperatures, the situation is similar as with GPU temperatures, only with slightly lower values of temperatures. When running in passive mode, recorded values of temperatures of supplying regulators shows identical problem, as it was with Accelero S1. Regarding MOSFET, this cooler is possible to be used only for not overclocked graphic cards. Performance with 92 mm fan at 800 rpm., is again problematic and rather not usable.