Page 1:Benchmarking Battlefield 1 in DirectX 12
Page 2:Mainstream PC, 1920x1080
Page 3:Mainstream PC, 2560x1440
Page 4:High-End PC, 1920x1080
Page 5:High-End PC, 2560x1440
Page 6:High-End PC, 3840x2160
Page 7:RX 480 And GTX 1060: Mainstream Vs. High-End
Page 8:Scaling: CPU Core Count
High-End PC, 2560x1440
Low Quality Preset
Our 2560x1440 results compare nine different cards from AMD and nine others from Nvidia. Even then, we had to cut some of the cards we benchmarked. There are just a ton of viable options for the range of quality settings at this resolution that might make sense with a Core i7-6700K-based platform.
Dropping all the way to Low quality, a Titan X continues to peg Battlefield 1’s 200 FPS ceiling with GeForce GTX 1080 not far behind.
Notice that the GeForce cards’ minimum frame rates are, at most, 20 FPS or so behind the averages. Flip over to the frame rate over time chart, and you’ll see relatively consistent trend lines. We’ll compare those results to AMD’s Radeons in the next set of charts because they differ from each other quite a bit.
In any case, even a GeForce GTX 780 Ti averages over 100 FPS at QHD. Playability isn’t an issue on a fast card using this game’s Low preset.
AMD’s last few generations also kick back average frame rates in excess of 100 FPS. But notice their minimums are all over the place. The average frame rate over time chart shows the HBM-equipped Fury X and Fury launching right into our benchmark sequence without missing a beat. But the boards with GDDR5 take a few seconds to get up to speed. The only exception is Radeon RX 470, which starts higher than several faster cards, but ultimately remains the chart’s slowest contender.
Certain events trigger frame time spikes across the Radeon line-up. We’re thinking these correspond to explosions during our run that “shake” the camera. The GeForce cards experience a similar phenomenon, though the effects don’t appear as pronounced or as frequent.
Medium Quality Preset
A step up to the Medium preset has a big impact on how Battlefield 1 looks, and it affects performance just as significantly. Even still, our slowest card, the GeForce GTX 780 Ti, keeps its nose above 60 FPS through the whole run. We’re waiting to see if the 780 Ti’s 3GB of memory becomes a more prominent bottleneck at higher detail settings or resolutions. But for now, a $500 GeForce GTX 1080 is almost exactly twice as fast as the 780 Ti, which sold for $700 three years ago.
More demanding detail settings smooth out the performance inconsistency most Radeon cards experienced during the first few seconds of our benchmark sequence.
Both Fiji-based boards enjoy a significant advantage over the rest of AMD’s portfolio. The Radeon R9 Fury X does battle with Nvidia’s GeForce GTX 1070, while the vanilla Fury is a bit faster than GeForce GTX 980 Ti.
Back in 2015, GeForce GTX 980 Ti was generally quicker than Radeon R9 Fury X. So it’s a pretty big deal that two years later, AMD beats that same card with its Radeon R9 Fury.
Meanwhile, the Radeon RX 480 8GB is quicker than GeForce GTX 980 and 1060 6GB.
High Quality Preset
Gamers sinking big bucks into high-end hardware want to see games the way their developers intended, with graphics quality options as high as possible. By simply dialing up to the High preset, rather than Medium, a mid-range card like GeForce GTX 1060 6GB sheds ~31% of its average frame rate, dropping from 84.5 to 58.1 FPS. With that said, even a GeForce GTX 780 Ti maintains >40 FPS through our run.
AMD’s fastest cards can’t compete with Nvidia’s, but the Radeon R9 Fury X does fare well against the GeForce GTX 1070. Unfortunately, Fiji-based boards are no longer readily available.
The Ellesmere-based Radeon RX 480 and 470 are, though. Both serve up playable performance at 2560x1440 using Battlefield 1’s High preset. Radeon RX 480 posts similar frame rates as a GeForce GTX 980, narrowly beating the 1060 6GB.
Our frame time over time charts show both HBM-equipped Fiji boards enduring small frame time spikes at regular intervals, similar to what we just saw from Nvidia’s GeForce GTX 780 Ti. These artifacts are interesting because they persist as we shift to Ultra quality, but are deemphasized as other influences cause much more significant frame time variance. Those aren’t the only 4GB cards we’re testing, so it’s not clear what imposes the evenly-spaced spikes.
Ultra Quality Preset
Our unevenness index makes the case that all of these GeForce cards serve up consistent-enough performance to be considered playable. The GeForce GTX 1060 6GB, 980 Ti, 970, and 780 Ti incur some fairly significant frame time spikes at similar points in our benchmark run. Incidentally, those are the cards at the bottom of the aforementioned index, which reflects smoothness.
The Fiji-based cards continue beating GeForce GTX 980 Ti, but the R9 Fury X now shows up behind GeForce GTX 1070.
Big frame time variance at the start of our run takes a toll on minimum frame rates and affects the smoothness metric, where five Radeon cards demonstrate noticeable stutter. Even still, most cards are playable.
MORE: Best Graphics Cards
MORE: All Graphics Content