Test System

Battlefield 1 Performance

T0 test the performance limits of Battlefield 1 and the newly minted Frostbite 3.5, we took to a 64-player server playing the Amiens map. It’s difficult without a built-in benchmarks to truly provide a consistent benchmarking experience, and a multiplayer-centric game such as this, performance can vary so greatly depending the action happening in any particular moment.


Battlefield 1

Using a DirectX 11 backbone we see that Pascal is indeed at the top of the list. The GTX 1080 and likewise the Pascal Titan X are providing very excellent performance and are certainly more than enough to provide a great experience. It’s unsurpassed, though that’s not to say that AMD’s offerings, and NVIDIA’s lower-end offerings re doing poorly here, either. Being a well optimized game engine, the lowest GPU that would be recommended at 1080P with ultra settings would be a GTX 960. Anything above that is more than fine to use attached to a decent processor.

Battlefield 1

Switching to DirectX 12 we see just how efficient AMD is with the next-generation API. The Fury X gains enough performance to outperform even the, admittedly older, Maxwell-based Titan X. Maxwell, as we know, isn’t very good when it comes to context switching between compute and rendering, so takes a small hit in performance in most cases. Such as here. Pascal, though, does find a small advantage by switching, though not nearly as prodigious as AMDs advantage. Realistically nearly all cards present can adequately play Battlefield 1 at Ultra settings, though the R7 370 may be a bit of a stretch, depending on how sensitive you are to frame rates.



When we turn up the resolution to 1440P we still see Battlefield 1 being playable on nearly anything. Provided, of course, you have a strong CPU to feed those hungry stream units and CUDA cores. Unsurprisingly the Pascal Titan X gives quite a bit of headroom and all but the GTX 960 and AMD R7 370 are unplayable. The R9 380X on up are are all able to give you a steady enough frame rate to actually enjoy your online experience on Ultra. Though SLI and now Crossfire, in the latest driver update from AMD, is supported, it’s hardly needed even at this resolution.


Switching to DX12 sees precisely what we expect. A small increase for Pascal, a slight decrease with Maxwell and a slightly larger increase with AMD’s offerings. The R7 370 still isn’t strictly playable, though the 380X sees a slightly higher average and even passes across 60 enough times to make it a much better experience. Fiji and Hawaii are rockstars, it seems, with driver improvements and DX12 providing just the boost they need to be more than viable in the modern age. Despite the age of Hawaii, the 390X remains quite the good choice, as does the Fury X, which now surpasses the Maxwell Titan X once again.


Battlefield 1

The Pascal Titan X was marketed as being a 4K capable card, and here it certainly does show that it’s possible. Furthermore, the R9 Fury on up also surprise with very smooth and capable experiences. The 1080, too, is 4K capable in Battlefield 1 though it seems to struggle just a tad, with a minimum that dropped during gameplay. There was a noticeable stutter, though I’m thinking it was an anomaly with our particular GTX 1080, though it was repeatable for us.Battlefield 1

Moving to DirectX 12 we see once again the Fury X even surpassing, though only slightly, the GTX 1070. The RX 480 could even be a candidate for 4K in a single card if you prefer the more cinematic feel given from lower framerates. It can actually add a measure of immersion, though in a fast-paced environment reliant on your ability to react, it just won’t do. Regardless, it isn’t a slideshow on anything but the GTX 960 and R7 370. DX12 proves that it is a worthwhile upgrade, though its effects are less so at higher resolutions due to it taxing the resources. The Pascal Titan X and GTX 1080 show, again, that 4K gaming is possible with a single card.