We continue a series of reviews with an analysis of the performance of the main games of the past 2019. We recently published a detailed test of graphics cards in Borderlands 3, now let’s talk about the latest game in the Gears of War franchise.
The game traditionally offers a rich story-driven adventure with the possibility of passing in co-op. Online modes are represented by co-op survival modes «Horde» and «Escape», where our team resists waves of bots. There are also classic multiplayer battles.
Gears 5 is based on the Unreal Engine 4. Visually, the game resembles the previous part, the effects and lighting have been tweaked a bit, the locations have become more diverse. There are destructible elements of the environment, which is actively used in the gameplay. And the story campaign periodically puts on small shows with explosions and massive destruction to highlight these features. In general, the picture is pleasing to the eye, and this is one of the most beautiful games on the UE4 engine.
The game only runs in the DirectX 12 API environment with support for new technologies such as tile resources and asynchronous computing. The latter technology affects performance, and in our review we decided to find out if it makes sense to activate it on AMD and NVIDIA graphics cards. Testing will take place in Ultra mode with different Async Compute settings.
Top solutions, mid-range video cards and budget models have been tested.
The most powerful graphics card is the GeForce RTX 2080 Ti Founders Edition. Recall that for older Turing, the Founders Edition option implies a slight boost in Boost.
There is the flagship NVIDIA of the old generation — GeForce GTX 1080 Ti in the reference version.
From AMD’s side, the most powerful solution at the moment is the Radeon VII.
Full list of participants:
All video cards were brought to the standard frequencies for their series. Acceleration was not used.
Characteristics of test participants
|Video adapter||GeForce RTX 2080 Ti FE||GeForce GTX 1080 Ti||GeForce RTX 2070 Super||GeForce RTX 2060 Super||GeForce GTX 1660||GeForce GTX 1650 Super||GeForce GTX 1060||GeForce GTX 1060||Radeon VII||Radeon RX 5700 XT||Radeon RX 5700||Radeon RX 5500 XT||Radeon RX 580||Radeon RX 480||Radeon RX 570|
|Core||TU102||GP102||TU104||TU106||TU116||TU116||GP106||GP106||Vega 20||Navi 10||Navi 10||Navi 14||Polaris 20||Polaris 10||Polaris 20|
|Number of transistors, million pieces||18600||12000||13600||10800||6600||6600||4400||4400||13200||10300||10300||6400||5700||5700||5700|
|Process technology, nm||12||16||12||12||12||12||16||16||7||7||7||7||14||14||14|
|Core area, sq. mm||754||471||545||445||284||284||200||200||331||251||251||158||232||232||232|
|Number of CUDA Stream Processors||4352||3584||2560||2176||1408||1280||1280||1280||3840||2560||2304||1408||2304||2304||2048|
|Number of texture blocks||272||224||160||136||88||80||80||80||240||160||144||88||144||144||128|
|Number of render units||88||88||64||64||48||32||48||48||64||64||64||32||32||32||32|
|Core frequency, MHz: Base-Boost||1350–1635||1480–1582||1605–1770||1470–1650||1530–1785||1530–1725||1506–1708||1506–1708||1400–1750||1605–1905||1465–1725||1607–1845||1257–1340||1120–1266||1168–1244|
|Memory bus, bit||352||352||256||256||192||128||192||192||4096||256||256||128||256||256||256|
|Memory frequency, MHz||14000||11008||14000||14000||8000||12000||8000||8000||2000||14000||14000||14000||8000||8000||7000|
|Memory size, GB||11||11||8||8||6||4||6||3||16||8||8||4||8||8192||4096|
|Supported version of DirectX||12 (12_1)||12 (12_1)||12 (12_1)||12 (12_1)||12 (12_1)||12 (12_1)||12 (12_1)||12 (12_1)||12 (12_1)||12 (12_1)||12 (12_1)||12 (12_1)||12 (12_1)||12 (12_0)||112 (12_0)|
|Interface||PCI-E 3.0||PCI-E 3.0||PCI-E 3.0||PCI-E 3.0||PCI-E 3.0||PCI-E 3.0||PCI-E 3.0||PCI-E 3.0||PCI-E 3.0||PCI-E 4.0||PCI-E 4.0||PCI-E 4.0||PCI-E 3.0||PCI-E 3.0||PCI-E 3.0|
The table shows the official specifications for GPU frequencies. The graphs show the full frequency range, including peak Boost values that are not specified in the specifications. This notation is used both for NVIDIA solutions and for new AMD models.
The test bench configuration is as follows:
- processor: Intel Core i7-6950X (email@example.com 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: Kingston SSDNow KC400 (512 GB, SATA 6Gb/s);
- secondary drive: WD Red WD30EFRX (3 TB, SATA 6Gb/s, 5400 rpm);
- power supply: Raidmax Cobra RX-800AE;
- monitor: ASUS PB278Q (2560×1440, 27″);
- operating system: Windows 10 Pro x64;
- Radeon Driver: AMD Adrenalin Edition 20.1.1/19.11.1;
- GeForce driver: NVIDIA GeForce 441.66/441.12.
The built-in benchmark is used, which was run six times on each video card.
The tests were carried out in three resolutions with the Ultra maximum quality profile selected. Disabled frame rate caps and the minimum fps setting that activates dynamic resolution. The graphs show the results of the minimum and average fps.
Activating asynchronous computing provides an increase in fps on AMD video cards by 2-3%, with the same modest drop in results for NVIDIA representatives. Acceptable results are shown by video cards starting with the Radeon RX 580 and GeForce GTX 1060. A little faster than the Radeon RX 5500 XT and GeForce GTX 1650 Super, which results in about 60 fps. The minimum difference between GeForce RTX 2070 Super and Radeon RX 5700 XT, especially if you compare the results in the optimal mode for each video card. The video memory load is approaching 5 GB.
As the resolution increases, the impact of asynchronous calculations is minimal, but the trend with increasing fps for AMD Radeon video cards remains. The GeForce RTX 2060 Super and Radeon RX 5700 perform well at 2560×1440, but more powerful graphics cards are needed if you want to go above 60 fps. At the same level, the results of the GeForce RTX 2070 Super, GeForce GTX 1080 Ti and Radeon VII are slightly weaker than the Radeon RX 5700 XT. It is worth noting that the old NVIDIA flagship looks the worst in the second mode, but it is still a small performance drop compared to the normal mode without Async Compute.
In the end, let’s compare the older video cards in high 4K resolution — 3840×2160.
The GeForce RTX 2070 Super outperforms the Radeon VII slightly without asynchronous computing, but after enabling them, the difference is minimal. These opponents give weak results, but a slight decrease in the settings will allow you to achieve optimal performance. The best and no alternative solution here is the GeForce RTX 2080 Ti. The video memory load can approach 7 GB.
Popular in the past, the GeForce GTX 1060 and Radeon RX 580 still cope with the game in Full HD resolution, but lose to the GeForce GTX 1650 Super and Radeon RX 5500 XT. The latter give an average result of 58-62 fps, and more powerful solutions will provide even better results and maximum comfort. Senior video cards GeForce GTX Super-series and models of the Radeon RX 5700 family will cope with a resolution of 2560×1440. 4K will only be mastered by the GeForce RTX 2080 Ti.
Wide graphics settings will allow you to achieve optimal performance on any modern PC configuration. The simplest solution for a little speedup is to activate the «minimum frame rate» setting, which will enable dynamic resolution, slightly degrade clarity in heavy scenes, but smooth out overall performance. On AMD Radeon, you can enable asynchronous computing — this will provide a couple of additional percentages of performance.