System power consumption
To test system power consumption, I stood in the entry hall of the chateau in Hitman's Paris level.
Under this load, the GTX 1070 Ti only needs moderately more power to run than the custom EVGA GTX 1070 we tapped to represent that card, and it uses a whopping 70W less than the RX Vega 56. Meanwhile, the RX Vega 64 still requires 148W more power than the GTX 1080 to offer roughly equivalent performance. Let's see how those extra watts translate into dBAs on my noise meter.
At idle, AMD's blower cooler and the semi-passive coolers on everything but the three Founders Edition cards are indistinguishable from the noise floor in my testing environment. The GTX 1070 Ti, GTX 1080, and GTX 1080 Ti all spin their fans at idle, but their noise levels are still hardly noticeable.
Load noise levels put the Founders Edition trio solidly in the middle of this grouping. Surprisingly, the GTX 1080 Ti FE doesn't get all that much louder than the GTX 1080 and GTX 1070 Ti FEs despite being asked to dissipate about 100W more heat. Despite its relatively high noise levels compared to the large dual-fan custom cards on the GTX 1070 and RX 580 I tested, the Founders Edition cooler is hardly unpleasant to listen to as blowers go. The noise it produces is a moderately high-pitched but broad-spectrum hiss.
Meanwhile, AMD's RX Vega blower cooler lands at the back of the pack in absolute noise levels. Like the Founders Edition blower, though, the RX Vegas' axial fan is a high-quality one, and it produces a fairly broad-spectrum noise. I was honestly surprised that both cards registered as high as they did on my sound meter. Still, there's no denying that the RX Vega 56 will be slightly louder than even the Founders Edition cards in use, and the RX Vega 64 will unquestionably make itself known without a closed and well-damped case.