Fashion for futuristic shooters continues. But DICE and EA decided to deviate from the general trend by releasing Battlefield. The familiar shooter in the unusual surroundings of the First World War sparkled with new colors and once again captured the attention of all fans of virtual battles.

The game traditionally pleases with a beautiful picture and epic battle scenes in large locations. A high level of detail with excellent elaboration of details is combined with good lighting and rich post-effects. HBAO+ progressive background shading enabled.

Visually, the game is varied. She is equally good at mountain landscapes and battlefields buried in mud with puddles and charred trees.

Battlefield 1 uses the Frostbite Engine 3 and supports both DirectX 11 and DirectX 12. The second mode is only available on Windows 10. The difference in picture quality is imperceptible, there are no new effects or new options in the settings menu. Actually, visual changes in the new DirectX were not announced. This is just an alternative rendering mode for newer graphics cards.

What version of DirectX is optimal? We will try to answer this question in this article. At the same time, we will conduct a comparative test of graphics accelerators of different levels and find out what class of devices can provide comfortable performance at different resolutions.

Testing was carried out by replaying a small game episode. After completing the first missions, it was decided to stop at the Cape Helles mission, where the assault on Gallipoli is played out. Soldiers land on the beach under heavy artillery fire, and with heavy explosions, performance sags more than in most other moments in the game.

This gaming episode can be called a stress test for video cards based on Battlefield 1. If we talk about multiplayer battles, then in network mode the graphics load is lower, so the frame rate is higher (with a sufficiently powerful CPU). But this test will reveal the maximum difference between video cards.

A short test sequence of actions between two control points was selected, which was repeated 7 times for each video card. Additional repetitions were performed if necessary. In DirectX 11 the frame rate was measured by Fraps, for DirectX 12 the Mirillis Action utility was used!

The testing included current mid-range solutions, top models from NVIDIA and a couple of representatives of the budget level.

The list of tested video cards is as follows:

All video cards have been tested at nominal and overclocked. An exception is made only for the GeForce GTX 1080 — there are no alternatives to the new NVIDIA flagship, it is the fastest by default. All participants work at the recommended frequencies, for which, if necessary, an appropriate correction was carried out. There is one exception — the GeForce GTX 980 Ti from ASUS. This adapter has a droop limit, resulting in 30 MHz overclocking, plus good cooling keeps Boost at its maximum level. Also, a new GeForce GTX 1050 Ti video card has been added to testing, a review of which will be released in the coming days.

test stand

The test bench configuration is as follows:

• processor: Intel Core i7-6950X (3.0@4.1 GHz);
• cooler: Noctua NH-D15 (two NF-A15 PWM fans, 140 mm, 1300 rpm);
• motherboard: MSI X99S MPower;
• memory: G.Skill F4-3200C14Q-32GTZ (4×8 GB, DDR4-3200, CL14-14-14-35);
• system disk: Intel SSD 520 Series 240GB (240 GB, SATA 6Gb/s);
• secondary drive: Hitachi HDS721010CLA332 (1 TB, SATA 3Gb/s, 7200 rpm);
• power supply: Seasonic SS-750KM (750 W);
• monitor: ASUS PB278Q (2560×1440, 27″);

• operating system: Windows 10 Pro x64;
• GeForce driver: NVIDIA GeForce 375.57;
• Radeon driver: AMD Crimson 16.10.2.

Test results

Let’s start with testing at 1920×1080 with DirectX 11’s Ultra graphics quality set to the highest quality. Additionally, the video memory limit is turned off. Note also that the Ultra mode offers TAA anti-aliasing by default.

Let’s look at the results in DirectX 11 first.

In Full HD, the GeForce GTX 1080 video adapter delivers over 100 fps even in a heavy test scene. The gap from the GeForce GTX 980 Ti is under 30%, and this despite the fact that the predecessor in this case is several percent faster than the reference. The simple GeForce GTX 1060 video adapter loses about 25% to our version of the GeForce GTX 980 Ti. The Radeon RX 480 is literally a couple of percent behind it. In overclocking, the advantage of the GeForce GTX 1060 over the Radeon RX 480 is slightly greater. The Radeon RX 470 and Radeon R9 290 have similar performance. At the bottom of the ranking is the GeForce GTX 960 with 2 GB of memory and the Radeon R9 270X. The new GeForce GTX 1050 Ti is up to 24% faster than the GeForce GTX 960, but part of that massive advantage comes from its larger 4GB memory. The game downloads over 3 GB in the test scene, downloads may be even higher in other locations. But in general, 4 GB of VRAM is enough for Ultra quality at 1920×1080.

Now let’s look at the performance situation with exactly the same graphics settings in DirectX 12.

The results are sudden and discouraging for NVIDIA representatives. This even forced us to repeat the tests, which once again confirmed the new alignment of forces. The Radeon RX 470 turns out to be a direct competitor for the GeForce GTX 1060, and the Radeon RX 480 nominally competes with the GeForce GTX 980 Ti. Activating the new DirectX drastically reduces the performance of older GeForces. The GeForce GTX 1080 and GeForce GTX 980 Ti lose more than 30%, for the GeForce GTX 1060 the performance drop is about 24%. At the same time, on the GeForce GTX 1050 Ti, the drop is about 13%. The trend is strange. AMD solutions have minimal performance losses, only a few percent. And the Radeon RX 480 and Radeon RX 470 do not even sag the minimum fps. Also, the new DirectX equalizes the GeForce GTX 960 and the Radeon R9 270X. However, this mode is still too tough for budget representatives; for them, in DirectX 11, you need to lower the graphics settings.

Now let’s test video adapters in 2560×1440 resolution. The younger participants are no longer included in this comparison.

In DirectX 11, the mid-range confrontation is won by the GeForce GTX 1060, but the Radeon RX 480 is slightly behind. About 30% faster than the GeForce GTX 980 Ti, and the GeForce GTX 1080 beats the old flagship by 27%. The Radeon RX 470 is inferior to the Radeon RX 480 by about 17% and loses up to 3% to the Radeon R9 290. In overclocking, both comrades are as close as possible to the performance of the Radeon RX 480.

Let’s see what will change under DirectX 12.

Once again, the balance is changing in favor of AMD solutions, but not so dramatically. The performance drop of the top GeForces is a few percent lower. As a result, the GeForce GTX 980 Ti manages to maintain an advantage over the Radeon RX 480. The GeForce GTX 1060 is slightly faster than the Radeon RX 470, lagging behind the AMD competitor by only 5% in nominal terms and less than 7% in overclocking. The transition from DirectX 11 to DirectX 12 has little effect on all Radeon graphics accelerators.

The potential of the GeForce GTX 1080 allows us to hope for comfort in the game even in 4K mode (3840×2160). Especially for this flagship video adapter, testing was carried out in this mode with different versions of DirectX.

The results of the GeForce GTX 1080 in 4K are quite good. And this, we recall, in a severe stress test. The inclusion of DirectX 12, of course, negatively affects the performance, but the decrease in fps this time is only 17%. It turns out that the difference between DirectX 11 and DirectX 12 decreases as the load increases and the initial fps level decreases.

conclusions

It’s time to take stock. If you plan to play at 1920×1080 resolution with maximum graphics quality, then you need a video card not lower than the GeForce GTX 1050 Ti. Theoretically, the GeForce GTX 960 with 4 GB of memory and increased frequencies can also cope with this task. The Radeon RX 470 and Radeon R9 290 will already deliver 60 fps performance. Not far from them, the regular version of the GeForce GTX 970 should go. In the confrontation between the Radeon RX 480 and the GeForce GTX 1060 with DirectX 11, the NVIDIA graphics accelerator has a slight advantage. But activating DirectX 12 dramatically changes everything — this mode has a minimal effect on Radeon performance and leads to a serious decrease in GeForce performance. Testing the GeForce GTX 1080 at different resolutions showed that this drop in performance decreases with increasing load, and the minimum difference is observed at heavy 4K resolution. Perhaps the new DirectX is not using GeForce resources efficiently, and a software update will gradually change this situation. For NVIDIA’s top solutions in the usual 1920×1080 resolution, such a drop in fps is not a problem, since the final fps is still high. However, it is highly recommended to enable DirectX 11 in the settings. In this case, the GeForce GTX 1060 will easily provide high fps at 1920×1080 and 2560×1440. This mode can also be recommended for the Radeon R9 290/390 line, although they have small losses in DirectX 12. Minimal differences between performance in different versions of DirectX show the new AMD Polaris solutions. The Radeon RX 480 and Radeon RX 470 are optimal graphics cards for Battlefield 1 in DirectX 12 mode, but they do not have any noticeable acceleration. So the very relevance of DirectX 12 for this game is in question, as in many other projects.

A little smoothing advice. Battlefield 1 supports TAA and FXAA. The first option well smooths out the stepped ladders on the edges of objects, but reduces the clarity of the picture. If you experience some discomfort at Ultra quality, simply select FXAA or discard any anti-aliasing and enable scaling using the initial rendering at a higher resolution (the «resolution scale» parameter). Even on a 1920×1080 monitor, you can use a working resolution of 2560×1440, if, of course, the performance of your video card allows it.