Radeon RX Vega 64 And RX Vega 56 Review: AMD Back In High-End Graphics

We also spent a little time tuning and checking power numbers with these new Radeon RX Vega cards, to see what kind of additional performance we could squeeze out of them. Like previous-gen Polaris-based Radeon cards, when boosting, frequencies and voltages scale upwards (power and temperature permitting) based on the GPU's workload at the time. With the WattMan tool built into AMD's drivers, however, users have the ability to manually alter frequencies and voltages on a per-voltage step (or state) basis and the power targets can be changed as well. Wattman also includes a new preset slider for Power Saver, Balanced, Turbo and Custom settings on Vega.

Unfortunately, due to time constraints and the fact that overclocking with Vega right now is limited -- along with a new driver drop hitting just hours before launch that enabled some level of consistent tuning -- we didn't get enough time to explore the overclocking and power ranges of Vega. As such, we were only able to experiment with AMD's new Wattman Performance per Watt profile slider, adjusting between the default Balanced mode, as well as Turbo, and Power Save.

WattMan Running On The Radeon RX Vega 64

Wattman Running On The Radeon RX Vega 56

Balanced, Turbo, And Power Save - Finding The Sweet Spot

On the Turbo setting, the Radeon RX Vega 64 picks up about a 2 - 3 percent gain in performance, at least in these tests and game engines, though mileage may vary with other games, resolutions, or image quality settings. Vega 56 picks up a bit more, with seemingly more thermal headroom with a 3 - 4 percent gain at the Turbo setting. Power Save, on the other hand, may be where it's at for many folks looking to tame acoustics and power draw, which we'll look at in more detail next. In Power Save mode, the performance bleed-off is less than 10 percent for Vega 64, but only about 2 percent for Vega 56. Third party AIB cards will likely offer different cooling solutions as well, so there might be even more headroom in the various types of designs that come to market, including AMD's own Radeon RX Vega 64 Liquid Cooled Edition. We're expecting that card in, by the way, in a day or so, so stick around.

Before bringing this article to a close, we'll cover a couple of final data points for power consumption and acoustics. Throughout all of our benchmarking and testing, we monitored noise output and tracked how much power our test system was consuming using a power meter. Our goal was to give you an idea of how much power each configuration used while idle and also while under a heavy gaming workload. Please keep in mind that we were testing total system power consumption at the outlet here, not the power being drawn by the graphics cards alone. It's a relative measurement that does give you a decent view of how much additional, or less power draw a graphics card is placing on a system.

As the saying goes, a picture is worth a thousand words. Radeon RX Vega 64, at its default, out-the-box Balanced mode setting draws nearly 40 Watts more power under load than a GeForce GTX 1080 Ti, a card that carries 3MB of additional frame buffer memory and delivers a lot more graphics horsepower. The Radeon Vega 56 is a lot more power-efficient and as you can see, it actually drew less power than the system config with the AMD Fiji-based Radeon R9 Fury X, a card that carries 4GB less HBM frame buffer and is some 15 - 20 percent slower. This speaks to Vega's performance-per-watt gains with generational improvements for AMD, but unfortunately, compared to NVIDIA GeForce 10 silicon, Vega still looks voraciously power-hungry.

Of note, on Power Save mode, there are some really nice power savings (and nicer acoustics to be had) without much performance sacrifice. If you scan the Power Save frame rates and scores again above, and do the math, you'll see it's worth it.

Radeon RX Vega Temps And Acoustics -
It's Not Quite Fermi But...

There's really no other way to state this, so we're just going to come out with it. Compared to competitive NVIDIA GeForce 10 cards in the same segment or even NVIDIA's most powerful consumer cards, the Radeon RX Vega 64 and RX Vega 56 are loud. They're not crazy, hair dryer loud, but louder than we're used to hearing lately, whether it be versus a GTX 1080 Ti or a Radeon RX 580 strapped in a pushing the pixels. On average, both cards push about 50db (Vega 56) to 53db (Vega 64) on an open air test bench with a decibel meter positioned about about 6 - 8 inches away from the back of the cards. Drop these cards into a case and things get quieter for sure, but their fans will tend to stay in the higher ranges with more challenging thermal constraints. The Power Saver setting on either card can pull them down another 2 - 4db, in addition to taming thermal build-up in general. Turbo settings do what you might expect, things get a little (not a lot) louder.

Thermally, the cards also get pretty warm, with Vega 64 measuring about 150ºF on its backplate at its hottest spot and Vega 56 toasting at about 140ºF. These temps aren't too alarming, however. And again, monitoring GPU temps, we observed relatively nominal core temperatures. Still, these cards do push a bunch of heat out from their rear IO plate when stressed. 

In short, Radeon RX Vega, in its current incarnation, needs air flow and is simply louder than anything we've tested in the same performance category currently. We wouldn't say it's obnoxiously hot or loud, but yes, it's a bit of a feisty beast.