Thermal and Noise Performance

CPU cooler reviews are handled by someone else on TechPowerUp, but since I was testing this kit under the watercooling category, I had no choice but to also use my own CPU (and GPU) cooling results from elsewhere to provide comparison results. But before we get to that, let us first see how the EKWB Fluid Gaming 240G performed by itself.

Please note that all of these results are obtained by testing coolers as they come out of the box. I did thermal tests on two different systems:

8 core: Intel i7 5960x (4.2 GHz at 1.2 Vcore) with quad channel DDR4 memory
(a) Fans and pump at 50% (Test 1)
(b) Fans and pump at 100% (Test 2)

4 core: Intel i7 4770k (4.5 GHz at 1.25 Vcore) with dual channel DDR3 memory
(a) Fans and pump at 50% (Test 3)
(b) Fans and pump at 100% (Test 4)

Here, 50% and 100% refer to the PWM duty cycle given the kit expects the end user to control all three PWM devices by using a single PWM control header. Each test was done in a climate-controlled hotbox at 25 °C ambient, but the results will still be reported in terms of a delta T (component – ambient) in °C. The ambient temperature should nearly be the same throughout the box, with a +/- 0.25 °C variability at most, though the ambient temperature sensor was still held in front of the intake fan(s) as should in my opinion be the case with CPU coolers. Each test/scenario was repeated twice, and the average value was taken to account for possible issues with mounting and TIM. I used my reference TIM (Gelid GC Extreme) for these tests since a lot of coolers only come with a single TIM application, and cure time was accounted for by measuring values after a steady state was achieved post burning in with an Intel XTU run. Aida64 was used to measure core temperatures.


There is a decent amount of scaling of cooling performance by going from 50% to 100%, with a lot having to do with the fans at 50% simply not pushing enough air to dissipate a lot of heat from the radiator. These numbers are fairly good, as we will see soon, but before we do, let us take a look at the cooler's noise levels as well. This was done with the cooler alone in an anechoic chamber at 19 dBA and a sound probe 6" away from the cooler, such that the fans and pump were equidistant as best as possible from the probe. No other noise source was inside the chamber.


The near-linear increase in noise levels is a result of a near-linear RPM increase of the fans' PWM duty cycle and the pump itself being outpowered by the fans at every step as far as overall noise goes. Note also that actual perceived noise will be quite lower than the values above when inside a closed case located a few feet away, and yet the cooler was very quiet at 25%-50% and easily bearable at 75% PWM duty cycle when tested in an open bed after unboxing. It is only at 100% speeds where there is a good amount of airflow noise, and I would say users would be happy with the cooler at 50%-75% more often than not.

I chose the results from Test 4 (max cooling performance) as a point of comparison to other CPU coolers, and I also have noise levels for the coolers with everything set to 100%.


This is where I was impressed the most - the aluminum everywhere was not as big a bottleneck as I had expected. For those curious, I also have some limited i7 5960X cooler numbers, and the trends are mostly in line there too. The EKWB Fluid Gaming 240G is within 2.5 °C of the best-performing copper metal 240 mm liquid coolers and in the top third of CPU cooler performance. Things are not as favorable as far as noise levels go, but as with most expandable AIOs, this is mitigated by slightly slowing down the fans and pump for a big decrease in noise levels and a minor decrease in cooling performance. Overall, if you are looking for just a CPU cooler, air coolers are still the way to go from a cooling/noise/cost perspective.

But using the Fluid Gaming 240G merely as a CPU cooler makes little sense. This is a full watercooling kit after all, so the least we can do is figure out how it performs with the GPU block in the picture as well. To do so, I introduced the NVIDIA GTX 1080 into the i7-4770K loop using an EK-FC GTX 1080 (nickel-plated copper) waterblock with every other expandable AIO cooler but the Fluid Gaming 240G, for which I used the EK-AC GTX Pascal. My expectations were relatively worse performance for the latter owing to the addition of another heat-transfer bottleneck in the aluminum GPU block, but the actual differences were what I was looking for here. The GTX 1080 was overclocked to 2 GHz on the core at 1.2 Vcore, and Unigine Heaven 4.0 was used to put 100% load on the GPU. Once again, everything was placed in our environmental chamber and two measurements were taken, with the average values reported below for both the CPU and GPU as the delta T (component - ambient).


Things went mostly as expected here, with the GPU being impacted the most by the swap to aluminum, even among the 240 mm AIOs. But something did change for the CPU - the weaker pump in the Alphacool Eisbaer did not handle the additional coolant flow restriction from the GPU waterblock well, and the relatively worse CPU block from Alphacool also did not help here, which had the Fluid Gaming 240G be ever so slightly better at CPU cooling in a CPU and GPU loop as compared to a copper metal AIO watercooler based off custom loop components. Now that is something I never, ever expected to see.