At the end of 2017, Bethesda Softworks and MachineGames presented a new game about the brutal agent Blaskowitz. The shooter Wolfenstein II: The New Colossus delighted fans with peppy gunfights. The game is also notable for the fact that it supports the Vulkan API. Therefore, it is interesting to see which video cards will provide the best performance in this API, and which solutions allow you to comfortably play at high or maximum graphics quality. Let’s find out in this article, where video cards from AMD and NVIDIA will be tested in different modes.

Wolfenstein II: The New Colossus is based on the id Tech 6 engine. Relative to the previous parts, the detail has grown, more saturated volumetric lighting effects.

In places, the picture looks washed out and blurry due to aggressive post-processing, but a little tweaking of the settings will improve clarity. More on this will be discussed below.

Solutions from the top, middle and budget segments, including new generation models and old video cards, will take part in comparative testing.

NVIDIA’s most powerful solution will be the GeForce GTX 1080 Ti.

Next comes the ASUS ROG Strix GeForce GTX 1070 Ti Advanced edition graphics card, which is identical to other models in this series in terms of frequency characteristics. You will learn more about the possibilities of cooling and overclocking from the corresponding review.

From AMD’s side, the standard-design Radeon RX Vega 64 will be the top choice.

The full list of participants is as follows:

All video adapters were brought to standard frequencies to match the reference options. The Radeon RX Vega 64 was tested with the stock BIOS in normal balanced mode (default settings). The Radeon R9 270X replaces the predecessor Radeon HD 7870. At standard quality settings, participants were tested at nominal and overclocked, additional tests with a change in the texture caching level were carried out only at nominal.

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 (Intel X99);
• memory: G.Skill F4-3200C14Q-32GTZ (4×8 GB, DDR4-3200, CL14-14-14-35);
• system disk: Kingston SSDNow KC400 (512 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 388.71;
• Radeon Driver: AMD Crimson Edition 17.12.2.

For testing, an episode with a shootout was chosen to simulate a typical game situation. We grab a laser gun and deal with a group of enemies, making several explosions along the way, after which firing from a machine gun is performed.

Performance measured using MSI Afterburner. Since the results on some video cards turned out to be unstable, the testing was carried out in two runs with several repetitions each time. For AMD video adapters, rough rounding was used for the minimum fps to integer values, and some error is possible there.

First, let’s briefly define the testing modes. The game has several graphics settings profiles:

• Mein Leben! (Mein Leben!).
• Uber (Uber).
• Super High (Very High).
• High (High).
• Average (Medium).
• Low

Tests will be conducted at settings from maximum to high. At high fps, there is a certain instability of the results, and video adapters with a small amount of memory show a slide show. Therefore, there was no point in testing everyone together. Participants are divided into several groups, and each is tested in different modes. Performed additional tests with a change in the «Image Streaming» parameter. It determines the level of texture caching and in some cases greatly affects performance.

Let’s start by testing at maximum quality at a resolution of 1920×1080. Standard settings Mein Leben! This is illustrated below with Image Streaming highlighted.

Results next.

The first surprise is that the GeForce GTX 780 Ti can barely cope with this mode, and the old video adapter has a huge lag behind the GeForce GTX 1060. Our 4 GB versions of the Radeon RX 480 and Radeon RX 580 are inferior to the GeForce GTX 1060, the biggest difference in minimum fps , the gain from overclocking is extremely weak. The Radeon RX Vega 64 shows a big advantage over the GeForce GTX 1070 Ti in terms of average frame rates, but the AMD leader has a lower minimum fps.

Now reduce the item «Image Streaming» to a minimum. This reduces texture loading. Initially, up to 5.3 GB was used, after lowering the parameter — 3.3–3.6 GB. Testing was conducted only for video cards with memory less than 8 GB, without overclocking.

Incredible performance boost for the Radeon RX 480/580. Our versions were clearly constrained by the amount of memory, and now they have revealed their potential, immediately overtaking the GeForce GTX 1060. However, this advantage only applies to the average frame rate, at some points the minimum fps is still lower. The GeForce GTX 780 Ti shows a relatively small boost, probably 3 GB is still a limiting factor.

Let’s lower the graphics settings to the level of Uber. With standard settings, tests at nominal and overclocked.

A slight increase in performance on older cards and a more significant increase on the Radeon RX 480/580. But in general, the situation is such that the Radeon RX 580 and GeForce GTX 1060 show the same average fps, and the minimum on AMD is lower. A serious advantage of the Radeon RX Vega 64 over the GeForce GTX 1070 Ti in terms of the average gaming performance level and the usual lag in the minimum fps.

Additional testing was carried out in the face value at a low level of Image Streaming.

And again, the Radeon RX 480/580 gets a serious acceleration with a minimal change in the results of the GeForce GTX 1060. The old GeForce GTX 780 Ti increases fps by 20%. The memory load drops from 4.7 GB to 3.3 GB.

Going to 2560×1440 resolution with Uber settings. Apply overclocking to GeForce GTX 1070 Ti and Radeon RX Vega 64.

The GeForce GTX 1060 has good results. But the Radeon RX 480/580 with 4 GB memory fades into the background, the minimum fps drops to the bar of 30 frames. Extremely low performance for the GeForce GTX 780 Ti. The ratio between GeForce GTX 1070 Ti and Radeon RX Vega 64 is standard, AMD has a good advantage only in average fps.

Additional low caching testing performed on video cards with less than 8 GB of memory.

And again, a cardinal acceleration of the Radeon RX 480/580 with a small amount of memory. In terms of downloads, we initially used about 5GB at 2560×1440, with a minimum Image Streaming of only 3.6GB.

Increasing the settings further had a negative effect on mid-range graphics cards, especially on the stability of the final data. Therefore, the comparison at 2560×1440 at maximum quality was carried out only for top solutions, the GeForce GTX 1080 Ti video adapter was added

In this mode, changing Image Streaming no longer plays a key role, although it will provide a few additional percentages of performance. The memory load in the test scene was initially 5.7 GB, which is still far from the maximum video buffer capacity of older graphics accelerators. The GeForce GTX 1080 Ti is in the lead, but pay attention to how little the Radeon RX Vega 64 is behind. Unfortunately, AMD’s representative shows the worst result in terms of minimum fps.

Let’s compare the top three at 3840×2160 at maximum quality.

All video adapters showed good results, allowing you to play in 4K. 8 GB of video memory is quite enough for this mode, so there is no drastic change in the results in the second graph. Maximum comfort will provide GeForce GTX 1080 Ti. From the experience of practical use, I note that with this video adapter you will observe about 60 fps only in heavy scenes with explosions or with dense volumetric lighting, in simple corridor locations it can be under 80 fps. A modest 7-8% is inferior to the Radeon RX Vega 64 in terms of average frame rates, but the minimum values ​​\u200b\u200bare traditionally the worst.

At the end of testing, let’s turn to budget accelerators. Their potential requires comparison at not the highest settings. Let’s test in «super high quality», additionally disable the motion blur effect, select SMAA anti-aliasing. A comparison was also made with low caching, but only at nominal frequencies.

The GeForce GTX 1050 Ti stands out here, thanks to 4 GB of memory. Video memory load is initially around 4 GB or higher, after reducing Image Streaming drops to 2.9 GB. All video adapters with 2 GB rest on memory. Therefore, even with an average performance of under 40 frames, there are drawdowns below 20. There is still a slight difference between the GeForce GTX 960 and the Radeon R9 270X or between the GeForce GTX 950 and the Radeon R9 270. The GeForce GTX 1050 Ti initially produces about 50 frames with a minimum of 30. The decrease in memory load provides a significant increase in the minimum fps, allowing us to talk about complete comfort when using this video adapter. It can be assumed that the Radeon RX 560 would have had good performance in the 4 GB version, but we did not have this accelerator on our hands.

We use a high quality configuration with all the edits of the previous configuration.

Minimal performance increase. Budget video adapters with 2 GB do not get rid of drawdowns. The balance between the GeForce GTX 960 and the Radeon R9 270X changes slightly, they are almost on the same level. Upload video memory up to 2.8 GB.

conclusions

Wolfenstein II: The New Colossus shows a nice picture with bright effects during gunfights. Graphics can be perceived ambiguously due to excessive blurring of the image. SMAA anti-aliasing will provide better clarity, although it does not neutralize ladders at the edges of objects worse. We also recommend turning off the motion blur effect. Less «soap» while reducing the quality of volumetric lighting. We strongly recommend lowering the value of the Image Streaming parameter to speed up video cards, especially with a small amount of memory.

System requirements are not very high, but the game is critical to the amount of video memory. Budget video cards with 2 GB can provide a good average fps at high quality, but there will be serious drawdowns and lags. Solutions with 4 GB feel much more confident. The GeForce GTX 1060 or Radeon RX 480/580 allow you to comfortably play at maximum quality (Mein Leben!) at 1920×1080, they can also play at 2560×1440 with not the highest settings. It is worth noting that the 4 GB Radeon RX 480/580 is worse than the GeForce GTX 1060, but only because of the memory. Reducing Image Streaming really speeds them up, and this shows that the potential of the Radeon RX 480/580 in the 8 GB version will be higher than that of the GeForce GTX 1060. With the GeForce GTX 1070 Ti, you can play in 4K. The GeForce GTX 1080 Ti and Radeon RX Vega 64 will provide even more comfort at high resolutions. At the same time, the gap between the Radeon RX Vega 64 and the NVIDIA flagship is very small when compared in terms of average gaming frame rates. With minimal fps, AMD has a strange situation, in skirmishes with explosions, the drawdown is greater than that of competitive solutions from NVIDIA. We tried to run tests on a different video driver, but it did not change the situation. If we talk about the Radeon RX Vega 64, then according to subjective impressions, the game runs perfectly, no slowdowns are noticeable. But with the Radeon RX 580 in heavy modes, lags are noticeable at the level of normal visual perception. That is, AMD Radeon has potential, it is high, but software polishing is required to achieve a more balanced performance without sudden fps drops.