The principal Tom’s Hardware graphics lab continues utilizing Cybenetics’ Powenetics hardware/software solution for accurately measuring power consumption.
Powenetics, In Depth
For a closer look at our U.S. lab’s power consumption measurement platform, check out Powenetics: A Better Way To Measure Power Draw for CPUs, GPUs & Storage.
In brief, Powenetics utilizes Tinkerforge Master Bricks, to which Voltage/Current bricklets are attached. The bricklets are installed between the load and power supply, and they monitor consumption through each of the modified PSU’s auxiliary power connectors and through the PCIe slot by way of a PCIe riser. Custom software logs the readings, allowing us to dial in a sampling rate, pull that data into Excel, and very accurately chart everything from average power across a benchmark run to instantaneous spikes.
The software is set up to log the power consumption of graphics cards, storage devices, and CPUs. However, we’re only using the bricklets relevant to graphics card testing. Sapphire's Nitro+ RX Vega 56 8G gets all of its power from the PCIe slot and two eight-pin PCIe connectors. Should it become necessary to retest one of the Limited Edition Radeon RX Vega cards with three auxiliary connectors, we can do that too.
If pricing and performance are Radeon RX Vega 56’s current strengths, then power is its most glaring weakness. The 12.5-billion-transistor Vega 10 GPU manufactured on GlobalFoundries’ 14LPP node is pushed hard to keep this card competitive, and its disadvantage is apparent right out of the gate when we measure idle power consumption.
Sitting there, doing nothing for 15 minutes on the Windows desktop, Sapphire’s Nitro+ RX Vega 56 8G averages almost 16W, peaking as high as 19W. Most of that power comes from the PCI Express slot, though the pair of eight-pin auxiliary connectors are continuously active as well.
Those idle numbers are downright tame compared to our measurements through three runs of the Metro: Last Light benchmark, though. Average power use in excess of 282W is more than 100W higher than thewe reviewed back in April. Obviously, any comparison between the two that involves efficiency is going to penalize AMD’s architecture.
Most of what Sapphire’s card needs is fed over the eight-pin auxiliary power connectors. Consumption over the PCIe slot’s +12V and +3.3V rails is much more conservative (and steady).
All four measurements combined yield the red line, which generally hovers under 300W, but does peak just over 313W in a worst-case scenario.
When we overlay the Nitro+ RX Vega 56 8G over other cards we’ve tested, Sapphire’s board appears to be in a completely different league. You’d never guess that the next contender down, in green, is another Radeon RX Vega 56.
GeForce GTX 1070 Founders Edition is slower across our performance benchmarks, but significantly more power-friendly. And that’s based on a previous-generation GPU.
At least Sapphire is deliberately conservative with current draw over the PCIe slot’s +12V rail. It stays under 2A through our benchmark sequence.
The Nitro+ RX Vega 56 8G’s average power consumption rises to 292W under FurMark.
Interestingly, though, this consistent workload imposes a slightly lower peak power measurement of 312.9W.
Charting power consumption over time confirms that the variance between all four lines is significantly reduced compared to what we saw under Metro: Last Light. The only downside to this is, again, a 10W-higher average.
All four of the cards in our comparison exhibit more consistent power consumption under FurMark. But whereas the two GeForces don’t have the headroom available to use more power than they did in Metro, both Radeon RX Vega 56es increase slightly.
Current draw over the PCIe slot increases a bit versus what we saw in a real-world gaming workload. However, it still doesn’t exceed 2A. Meanwhile, both eight-pin auxiliary connectors accommodate between 10 and 12A.
MORE: Best Graphics Cards
MORE: All Graphics Content