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 — this time at 1080p medium instead of our normal 1440p ultra, due to the weaker GPU. We also test with the FurMark stress test at 1600x900. Our power testing PC uses an open testbed, as that's required for all the extra wires and riser card, and it's the same Core i9-9900K that we've used for the past several years.
As noted earlier, we had to install a second graphics card in the PC to get the A380 to POST, but we disabled the card in Device Manager and our equipment doesn't include the power use from the other PCIe slot. That doesn't mean we're free and in the clear, however, as GPU-Z is used to collect data on GPU clocks, GPU utilization, and fan speeds. It seems to have worked, but the results were slightly more chaotic than we're used to seeing.
Average power use while running Metro Exodus was 73W, though peak power reached as high as 87W. While the GPU was under load during the Metro benchmark, power use ranged between about 65W and 85W. FurMark, on the other hand, generated a far more consistent load and kept the Gunnir card pegged at very close to its 95W TBP — 20W higher than Intel's reference clock target, thanks to the overclock.
Clock speeds during our Metro and FurMark tests were mostly flat, with the GPU cores sitting right near the 2450 MHz mark. While Intel says it rates the Arc GPUs by "average gaming clocks," either Gunnir is doing something different or the A380 can simply sit at the maximum 2450 MHz boost clock in most workloads. There was a slight amount of clock speed fluctuations with FurMark, but average clocks were only 12 MHz less than the maximum clock.
GPU temperatures were pretty flat, peaking at 64C and averaging 61C over the course of the five loops. As you'd expect, FurMark temperatures were a bit higher, hitting a steady plateau of 67C. Fan speed on the other hand was higher than we'd expect for such a large cooler, with dual fans no less. Peak fan speed was 2,067 RPM in Metro, though the average fan speed was quite a bit lower at just 1723 RPM. FurMark hit slightly higher peak speeds of 2,227 RPM.
While that's on the higher end of the RPM spectrum, particularly for a 95W graphics card, the Gunnir card still wasn't particularly loud. Unfortunately, due to the issues with getting the card running in our 9900K test PC (we had to install a second GPU), noise measurements were thrown off. We ended up testing in a different PC, one that has a higher noise floor of 37 dB(A) for our SPL meter position.
We positioned the SPL (sound pressure level) mic between the two fans at a distance of about 10cm to minimize the impact of the CPU cooling fans, and the Gunnir card created up to 49.1 dB(A) of noise while gaming, with some slight fluctuations. With the case side on, that dropped to 42 dB — not silent by any means, but also not terribly audible. We couldn't set a static fan speed due to the lack of support by GPU overclocking and tuning utilities, and GPU-Z didn't report fan speed as a percentage, so we're not sure how much louder the card could get.