Retested the cards with the new drivers.
March 1, 2010
The i3DSpeed project is dedicated to testing performance of a large number of graphics cards under the Windows OS.
Traditionally, we offer you performance diagrams of popular graphics cards and inform about the best price/performance deals in the market.
February 2010, Issue #26(115)
We'd like to clarify our selection of benchmarks. We tried more games than we're using now, including S.T.A.L.K.E.R.: Call of Prypyat and Batman: Arkham Asylum. Unfortunately, not every game is convenient for automated benchmarking and logging. Speaking of Call of Prypyat, the developers removed the previously available demo recording and playback features for some reason. Instead, they added an in-game benchmark which is very slow and, essentially, is not part of the game anymore. The sluggishness is the key drawback though. It's physically impossible to test 24 graphics cards in 4 resolutions and 2 modes, and meet the deadline. Batman wouldn't do for the same reason. Accessing its benchmark means forcing your way through numerous cutscenes which cannot be removed. Besides, it doesn't support logging of test results.
Thus, we have selected games with decent benchmarks which let us automate the tests. Note that we couldn't care less about which game belongs to which marketing program of NVIDIA or AMD. As said above, our only criterion is whether a game offers a good automated benchmark.
As for the two benchmarks from Unigine, we are grateful to their developers for helping us understand how they work. We decided to use them to the max (Heaven is used in two modes, for example). As more DX11 games with good benchmarks are released, we'll try to add them to our list of benchmarks, too.
- Intel Core i7-920 2667 MHz CPU
- ASUS P6T Deluxe motherboard on the Intel X58 chipset
- 3GB of 1066MHz DDR3 SDRAM from Corsair
- WD Caviar SE WD1600JD 160GB SATA HDD
- Tagan TG900-BZ 900W PSU
- Windows 7 Ultimate 32-bit, DirectX 11
- 30" Dell 3007WFP monitor
- NVIDIA ForceWare 196.34 beta
- ATI CATALYST 10.2
- Far Cry 2 (Ubisoft) -- DirectX 10.0, shaders 4.0 (HDR), built-in benchmarking utility (the Middle level), maximum settings.
- CRYSIS 1.2 (Crytek/EA) -- Rescue, DirectX 10.0, shaders 4.0, Very High settings, batch file, timedemo available by e-mail request.
- CRYSIS Warhead (Crytek/EA) -- Cargo, DirectX 10.0, shaders 4.0, Very High settings, batch file, timedemo available by e-mail request.
- Colin McRae: DiRT 2 (Codemasters) -- DirectX 10.0/11.0, Ultra High settings, run the benchmark as follows: "dirt2.exe -benchmark example_benchmark.xml".
- Warhammer 40.000: Dawn Of War II (Relic Entertainment/THQ) -- DirectX 10.0, Super High settings, run the benchmark from the in-game menu.
- Tropics Benchmark (Unigine) -- DirectX 10.0, High settings.
- Heaven Benchmark (Unigine) -- DirectX 10.0, High settings.
- Heaven Benchmark (Unigine) -- DirectX 11.0, High settings.
- 3DMark Vantage 1.01 (FutureMark) -- DirectX 10.0, shaders 4.0, multitexturing, Extreme settings.
- RightMark3D (iXBT.com) -- DirectX, a synthetic benchmark, Shaders 2.0/2.0b/3.0.
We'd like to thank Unigine for helping us set up their benchmarks.
In the brackets: ROP/TMU clock rate, shader unit clock rate, effective memory clock rate.
- ATI RADEON HD 4670 512MB GDDR3 1.0ns (750/750/2000 MHz)
- ATI RADEON HD 5670 512MB GDDR5 (775/775/4000 MHz)
- ATI RADEON HD 4850 1024MB GDDR3 1.0ns (625/625/2000 MHz)
- ATI RADEON HD 4870 1024MB GDDR5 (750/750/3600 MHz)
- ATI RADEON HD 5750 1024MB GDDR5 (700/700/4600 MHz)
- ATI RADEON HD 5770 1024MB GDDR5 (850/850/4800 MHz)
- ATI RADEON HD 4890 1024MB GDDR5 (850/850/3900 MHz)
- ATI RADEON HD 5850 1024MB GDDR5 (725/725/4000 MHz)
- ATI RADEON HD 5870 1024MB GDDR5 (850/850/4800 MHz)
- ATI RADEON HD 5970 2x1024MB GDDR5 (725/725/4000 MHz)
- ATI RADEON HD 5870 CrossFire 2x1024MB GDDR5 (2x850/850/4800 MHz)
- ATI RADEON HD 5970 CrossFire 2x2x1024MB GDDR5 (725/725/4000 MHz)
- NVIDIA GeForce GT 220 1024MB GDDR3 1.2ns (625/1380/1600 MHz)
- NVIDIA GeForce GT 240 1024MB GDDR5 (585/1418/3780 MHz)
- NVIDIA GeForce 9600 GT 512MB GDDR3 1.0ns (650/1625/1800 MHz)
- NVIDIA GeForce 9800 GT 512MB GDDR3 1.0ns (600/1512/1800 MHz)
- NVIDIA GeForce 9800 GTX+ 512MB GDDR3 0.8ns (736/1838/2200 MHz)
- NVIDIA GeForce GTS 250 1024MB GDDR3 0.8ns (756/1890/2300 MHz)
- NVIDIA GeForce GTX 260 (216sp) 896MB GDDR3 0.8ns (575/1240/2000 MHz)
- NVIDIA GeForce GTX 275 896MB GDDR3 0.8ns (633/1400/2260 MHz)
- NVIDIA GeForce GTX 285 1024MB GDDR3 0.8ns (650/1475/2500 MHz)
- NVIDIA GeForce GTX 295 2x896MB GDDR3 0.8ns (575/1240/2000 MHz)
- NVIDIA GeForce GTX 285 SLI 2x1024MB GDDR3 0.8ns (2x650/1475/2500 MHz)
- NVIDIA GeForce GTX 295 SLI 4x896MB GDDR3 0.8ns (4x575/1240/2000 MHz)
Archived graphics cards
Write a comment below. No registration needed!
AMD FX-8350 Processor
The first worthwhile Piledriver CPU.
i3DSpeed, March 2013
Added the test results of AMD Radeon HD 7850 1024MB, AMD Radeon HD 7790 (standard and overclocked), ASUS Ares II (Radeon 7970 GHz CrossFire), ASUS Ares II CrossFire, NVIDIA GeForce GTX 650 Ti Boost, GeForce GTX Titan. Replaced the 3DMark11 and Formula 1 (