For our power, temperature, clock speed, and fan speed testing, we use Powenetics testing hardware and software. We capture in-line GPU power consumption by collecting data while looping Metro Exodus at 1440p ultra and running the FurMark stress test at 1600x900. Our test PC remains the same old Core i9-9900K as we've used previously, to keep results consistent. We tested the Sapphire card in its default mode and also enabled AMD's Rage mode in the drivers, which applies a potential 10% power limit increase.
Starting with the Metro Exodus results, Rage mode increased power use by 22W (5.7%) and resulted in slightly higher performance. However, it's not something you'd notice, though neither is the power increase. FurMark showed a slightly larger increase in power use, but either way, the Sapphire card comes in well below the Asus RTX 3090 Ti. Does anyone spending $1,000 or more on a graphics card really worry about 50–100 extra watts while playing games? Probably not.
There's a huge difference in clock speeds between AMD's RDNA 2 GPUs and Nvidia's Ampere GPUs. The Asus 3090 Ti is the fastest card we've ever tested from Nvidia in terms of GPU clocks, just breaking the 2 GHz mark. Meanwhile, AMD's GPUs run in the 2.25–2.6 GHz range. Higher clocks can help in a variety of ways since they apply to the shader cores, ROPs, texturing units, Infinity Cache, etc. This is part of how AMD still manages to compete with Ampere, even though, on paper, Nvidia's GPUs should be quite a bit faster.
Compared to the reference RX 6900 XT, Sapphire's card clocked 260MHz faster at its default settings, and nearly 330MHz higher (14.6%) higher with Rage mode enabled. And Metro Exodus likely isn't hitting the latest GPUs quite as hard as it used to, so the gap might be even wider in some games. That's likely the case with those games that showed more than a 15% lead for the 6950 XT over the 6900 XT.
Clock speeds in FurMark were lower, but even there, the Sapphire card easily broke 2.2GHz and clocked on average 10% higher than the RX 6900 XT.
There's little surprise with the massive heatsink and fans on the Sapphire Nitro+ Pure, which provided good cooling. Rage mode actually caused the fans to spin faster, yielding a 6C drop in temperatures, but in either case, peak temperatures were well within the reasonable range.
You can see the fan speed increase in the other charts, where Rage mode boosted the fan RPMs by about 25% in Metro, with an even larger 35% increase in RPMs in FurMark. Noise levels are, of course, more important than the raw fan RPMs.
We measured noise levels at 10cm using an SPL (sound pressure level) meter. The SPL was aimed right at the middle GPU fan in order to minimize the impact of other fans like those on the CPU cooler. The noise floor of our test environment and equipment measures 33 dB(A). At stock settings, the fans on the Sapphire card were only spinning at an average of 26% (across all three fans, though the middle fan normally spins a bit faster). This resulted in a noise level of just 37.3 dB(A), one of the quietest results we've measured — only the Asus RTX 3070 Noctua did slightly better, by 1 dB. Rage mode bumped the fan speed up to 30% and 39.8 dB(A), which was still very good.
We also test with a static fan speed of 75% just to see how much headroom there might be for additional overclocking. As you'd expect, that resulted in substantially more noise, with the Sapphire card hitting 56.2 dB(A). That's comparable to what we've measured on many other third-party factory overclocked cards.