For a full-hardware workout, visit https://benchmarks.ul.com for our system warm-up and stress test of choice.
For benchmarking methodology please see our game benchmark method guide.
Test results are gathered and produced on CapFrameX. This makes it easier for use to get both line graph comparison and raw averages without extra tools. Simply the easiest tool for benchmarking and its available for everyone to use, free of charge. Check it out at capframex.com.
Since this is a GPU review, we benchmarked the area of the games that put heavy load on the GPU.
All our test runs are repeatable, click the links below for area and details. Read our benchmarking methodology.
- Counter-Strike: Global Offensive: FPS Benchmark Workshop Map
- DOTA2 – The International Main Event – Day 6: T1 vs PSG.LGD Game 2 (48:00) – The 10-Man Buyback Fight
- Rainbow Six: Siege – Benchmark Mode
- PUBG Battlegrounds – Custom Scene
- Call of Duty: Modern Warfare 2 (2022)
- Valorant
- Grand Theft Auto V – Palomino Highlands
- Shadow of the Tomb Raider – Kuwaq Yaqu
- Destiny 2 – The Tower
- Forza Horizon 5
- Elden Ring – Custom scene, FPS cap removed, offline
- Cyberpunk 2077 – Little China, noontime
- Marvel’s Spider-Man – Empire State area
- F1 2022 – Bahrain, night time
- Microsoft Flight Simulator – Landing mission, YSSY
See our Youtube playlist for benchmark sequences.
Note: Some proprietary technologies of NVIDIA like PCSS, HBAO+, and HairWorks work on AMD GPU’s BUT to maintain uniformity amongst GPUs, these have been turned OFF.
You can click on any of the benchmark charts enlarge. You can also move forward and backwards to quickly navigate through our charts via gallery view. For this test, only the out-of-box normal mode will be tested.
Kindly let me know if you spot an errors in the charts. I do my best to keep them error free but while test results are reliable and accurate, bringing them over to Excel and relying on formulas to generate the reports sometimes can cause mix-ups.
Notes:
- All data are gathered from exactly the same system, with exactly the configuration we list here. No data is reused from another system or from any variations of. We gather data from only one system as indicated here.
- Graphics cards are allowed to heat up prior to benchmarking. Cooler graphics cards may boost higher than normal.
- Following up on the above, we try to enjoy the game and play a bit before proceeding to the actual benchmark scenario. This allows us to detect any other problems like stuttering, frame skipping, or any other problems.
- Games that receive graphical updates that affect performance e.g. (DOTA2 moving from DX9 to DX11) will be retested completely.
Meh..mediocre