Fans & Noise
Fan Control & Fan Curves
To start, we wanted to measure the performance of all three fans.
The readings we took with our laser confirmed that what we observed in software more or less matched the actual fan speeds. So that we wouldn't have to work with three fan curves in one chart, we chose to graph their mean. This is accurate enough, since the two right-hand fans never spin at more than 100 RPM faster than the left fan, even under maximum thermal load.
Under a gaming workload, the difference between running on an open test bench and inside of a closed case is apparent. One reason for this is the unguided vertical airflow. Unfortunately, if the card is installed horizontally, some heated air can flow back under the card, at which point it's blown through again. If the exhaust was better-directed and guided towards the rear (the top of the card, once it's installed), you'd see lower temperature numbers. Our experiments confirmed this.
Since the GPU stays cooler during our stress test, the fans spin a bit slower as well.
We stumbled upon one rather puzzling behavior that we managed to reproduce only a few times and consider a bug. During the card's cool-down, the middle fan stops first. Now, if the card catches just the right (or actually, the wrong) moment to go back to full load, this fan remains inactive. Consequently, during gaming the card heats up to just above 80°C. At that point, only a reboot is able to get the middle fan working correctly again.
Noise
On the first page, you may have noticed a "Custom-Made Proprietary Measurement Chamber" in the test setup table. We're often asked about the minimum measurable noise level in that room. So, today we include a control reading for reference.
In our gaming loop and on an open test bench, we end up with almost 41 dB(A). That's acceptable for a two-slot GeForce GTX 1080 Ti armed with three fans. This goes to show, however, that slimming the board down does have consequences. Greater airflow ultimately has to compensate for reduced cooler surface area.
We observe almost no low-frequency bearing noise in the sound spectrum; this is rare, and incredibly pleasant. The measured 41 dB(A) is clearly audible, but the noise isn't obnoxious by any means.
In an extreme example, during our stress test with a maximum power target of 127%, fan noise may increase to just about 44 dB(A). This is very much on the far end of what is still reasonably acceptable.
EVGA's GeForce GTX 1080 Ti FTW3 Gaming sports a two-slot cooler that performs admirably, considering much of the competition relies on 2.5-slot designs. This thermal solution isn't super-quiet, but because the noise it generates isn't grating in any way, the outcome is still acceptable. When you install the GeForce GTX 1080 Ti FTW3 Gaming in a closed case, the more room you can leave between the card and chassis, the quieter those fans will be. Furthermore, try directing ambient airflow from the bottom of the case, below the card. This yields better results than unguided air coming from the front.
MORE: Best Graphics Cards
MORE: Desktop GPU Performance Hierarchy Table
MORE: All Graphics Content