Naturally, the memory runs on the XMP-II profile so that it reaches the full 7200 MHz. Additionally, we cranked the power target of the RTX 5080 all the way up to the maximum permitted 125%, and raised the Boost Clock setting from 2700 MHz default to 3000 MHz. We could squeeze more out, but this is just fine.
For the fan curve, Noctua offers a JSON file you can import into Fan Control, available for download here with a full explanation. We decided to download this fan curve and use the quietest of the three presets, but we did make a couple changes.
Article continues below 
By default, the curve sets up a virtual sensor that combines data from the CPU and GPU, and links this to the fan curve. However, this meant that we couldn’t quite apply the fan curve to all devices.
For the case fans, it works very well. But I also wanted to use the exact same curve on the CPU cooler and the GPU. So, I duplicated the quiet profile twice, and linked one of them to the GPU’s core temperature, and another to the CPU’s package temperature reading.
The curve has a hysteresis of 5% / 5 seconds, which ensures the fans don’t ramp up or down in speed too quickly upon brief temperature spikes. This keeps the system running calm and consistent.
It’s worth noting that when using Fan Control, you do have to disable other software that may control the fans. Because the Flux Pro has a little display for the CPU and GPU temperatures, you also need to have Antec’s iUnity installed, but it’s important to remember to turn its fan control feature off. I didn’t at first, and kept wondering why Fan Control wasn’t behaving right.
Test Results
Test | Duration/Score | CPU Temp | GPU Temp | dBA | System Power |
|---|---|---|---|---|---|
Sleep |
|
|
| 29.0 | 4 W |
Light Browsing |
| 37 | 28.2 | 29.0 | 108 W |
DXO-Export | 16:45 | 68 | 39.6 | 29.4 | 309 W |
3DMark Speedway | 9210 | 61 | 64.9 | 29.6 | 499 W |
Cyberpunk | 104 FPS | 76 | 62.8 | 30.3 | 527 W |
Furmark + CPU Burner |
| 97 | 70.6 | 41.2 | 832 W |
All Fans Full |
| 96 | 61.7 | 48.3 | 841 W |
Under light browsing, the system barely consumes more than 100 watts. In fact, when not doing anything at all, it hovers around 97, but for a fair test I measure over the course of 5 minutes doing basic desktop things like web browsing, chatting, and writing this document.
Exporting photos loads up the CPU nicely, and gives the GPU some work too. All remains incredibly quiet, barely increasing in noise level over the noise floor of the room.
3DMark Speedway and Cyberpunk 2077 are much more real-life loads that balance a burden on both the CPU and GPU. Power consumption for these two tests hovers around 500 watts, and noise levels remain whisper quiet. The fans do come in a little bit, and there is a smidgeon of coil whine from the GPU, but it’s not nearly as audible as in the open-air PC I borrowed the GPU from.
For the synthetic workload, I’ve changed to using Furmark together with its built-in CPU burner, as I found that Prime95 sometimes lead to inconsistencies when running at the same time as Furmark – it would hog too many resources, and Furmark’s built-in CPU burner seems to ensure the GPU’s workload remains unaffected.
With that tool, power consumption raises to a staggering 827 watts, which is significantly more than any real-world workload would ever place on the system. The PC was also no longer quiet at this point.
Lastly, with all the fans on full speed, the GPU runs almost 10 degrees cooler, but the noise penalty simply isn’t worth it.
I Forget I’m heating the Room
Now, what I’m about to say is going to sound ridiculous, but bear with me. When I’m writing, I can end up sitting still at my desk for hours. Gaming, in that respect, is far more active, but when writing, I get cold.
When I’m in the type of flow that has me walking around the house, I’ll just turn on the central heating that runs on gas. But gas has become quite expensive in the Netherlands, so what I’ll sometimes do is simply run Furmark to heat up only the room to save money.
With past systems, this would be a bit intrusive, but with this Noctua system, especially in the Flux Pro where I can barely hear the coil whine anymore, something quite funny has been happening: I forget Furmark is running in the background.
I’m not kidding – I’ve once had Furmark run for 17 hours straight, when I only wanted a bit of heat in the morning. I was at my desk most of the day, and at some point, late in the evening, it was getting too warm in the room. I looked at my taskbar to close all my apps for the night, and that’s when I realized, I’d left Furmark going all day, minimized – and I hadn’t noticed.
