We take it for granted that each successive generation of GPUs, and indeed all other computer components, outperform their immediate predecessors in net performance and - if not always - in performance-to-price ratio, which has obviously regressed in recent years.
However, a net increase in computing power over the best achievements of the past is the lot of flagship gas pedals, and more affordable solutions inevitably overlap with the models of the predecessor series.
In such cases, the results of the debut reviews of a new graphics card cannot be considered the last word.
As a rule, it should take some time before the GPU driver, and most importantly, games will be able to use the resources of the new architecture with maximum efficiency, and then we may have to look at the device in a new light.Fortunately, now we have an opportunity to do just that - to study how the performance of video cards of this or that family has changed over the long term.
GPU creators find it increasingly difficult to conquer new frontiers of computing power, most games use the same rendering tools since the release of DirectX 12 API, and only relatively recently stopped actively supporting such ancient gas pedals as GeForce GTX 700 series from NVIDIA and Radeon 300 from AMD.
Previously, we tried to find out how much influence on the performance of \"red\" and \"green\" graphics processors had driver optimization and came to the conclusion that video cards of all series, which were presented after the GeForce GTX 980 Ti and Radeon R9 390X, for the past time became more or less faster, and the greatest success by improving software, contrary to popular theory, achieved NVIDIA products (especially the GeForce 20 family).
It remains to consider the optimization from the other side - in games, which are focused on the continuously increasing GPU performance level and gradually master the rendering functions implemented in the hardware of the new architecture.
To do so, we took benchmark results from several generations of graphics cards in older games, the same ones used for driver testing, and compared them with what they can do in current games.⇡#Test bed, testing methodologyTest bedCUAMD Ryzen 9 5950X (4.4GHz, fixed clock speed all cores)MotherboardASUS ROG Strix X570-E Gaming (Resizable BAR incl.)RAMG Skill Trident Z RGB F4-3200C14D-16GTZR, 4 × 8GB (3600MT/s, CL17)ROMIntel SSD 760p, 2048GBPower supplyCorsair AX1200i, 1200WCooling system CPUCorsair iCUE H115i RGB PRO HTBasementOpen StandOperating SystemWindows 10 Pro x64GPU AMDRadeon R9 390XAMD Radeon Software Adrenalin 2020 Edition 22.
6.1 for Legacy ASICsOther graphics cardsAMD Radeon Software Adrenalin 2020 Edition 22.
6.1 GPU Software for NVIDIAGeForce GTX TITANNVIDIA GeForce Game Ready Driver 473.81Other Graphics CardsNVIDIA GeForce Game Ready Driver 516.40 Old GamesAPI GameAPI Testing MethodGraphics SettingsFull-screen Anti-AliasingDeus Ex: Mankind DividedDirectX 12Inbuilt BenchMark UltraTA graphics qualityGrand Theft Auto VDirectX 11Inbuilt BenchMax.
graphics qualityMSA 4xRise of the Tomb RaiderDirectX 12Built-in benchmarkGraphics quality Very HighTAATom Clancy's the DivisionDirectX 12Built-in benchmarkGraphics quality UltraTAATotal War: WARHAMMERDirectX 11Built-in benchmarkGraphics quality UltraFXAAThe Witcher 3: Wild HuntDirectX 11OCATGraphics quality Ultra, HighTAA postprocessing qualityNew gamesPlayAPI testing methodGraphics settingsFull-screen anti-aliasingAsasin's Creed ValhallaDirectX 12Inbuilt benchmarkMax.
graphics qualityTAA HighBorderlands 3DirectX 12Built-in benchmarkMax graphics qualityTAAASbyeProk 2077DirectX 12Built-in benchmarkMax graphics qualityTAADOOM EternalVulkanOCAT, Mars Corre mission startMax.
graphics qualityTSSAAFar Cry 6DirectX 12BinchmarkMax.
graphics qualityTAaAMetro ExodusDirectX 12BinchmarkMax.
graphics quality; Shading Rate 100%TAARed Dead Redemption 2VulkanBinchmarkMax.
graphics qualityTAA HighTotal War: WARHAMMER IIIDirectX 12Binchmark (Battle Benchmark)Max.
Graphic qualityTAA HighWatch Dogs: LegionDirectX 12Inbuilt BenchmarkMax graphic qualityTAAverage and minimum frame rates are derived from the array of individual rendering times for most test games, which is recorded by the built-in benchmark (or the OCAT utility if the benchmark is absent).
The minimum frame rate value corresponds to the 1st percentile of the distribution.
In case the built-in game benchmark does not provide data on individual frames, the conclusions of the benchmark itself are used.Old devices whose active support has already been discontinued (GeForce GTX TITAN and Radeon R9 390X) were tested under the latest available version
0 Comments