Testbed and Methods

All graphics cards were benchmarked in a closed system case with the following configuration:

• Mainboard: DFI LANPARTY DK X48-T2RS (Intel X48, LGA 775, BIOS 03/10/2008)
• CPU: Intel Core 2 Extreme QX9650 (3.0GHz, 1.15V, 2x6MB L2 cache, FSB: 333MHz x 4, Yorkfield C0)
• CPU cooler: ASUS Triton 88 (PWM-based control within 830-2200rpm)
• Thermal interface: Arctic Silver 5
• System memory:
  • 2 x 1024MB DDR2 Corsair Dominator TWIN2X2048-9136C5D (Spec: 1142MHz / 5-5-5-18 / 2.1V)
  • 2 x 1024MB DDR2 CSX DIABLO CSXO-XAC-1200-2GB-KIT (Spec: 1200MHz / 5-5-5-16 / 2.4V)
• Graphics cards:
  • Diamond Radeon HD 4890 GDDR5 (1024MB/256 bit, 925/4400MHz)
  • HIS Radeon HD 4870 X2 GDDR5 (2x1024MB/256 bit, 750/3600MHz)
• Disk subsystem: Western Digital Caviar Black WD6401AALS (SATA-II, 640GB, 7200rpm, 32MB cache, NCQ)
• Optical drive: Samsung SH-S183L DVD-burner
• System case: Thermaltake Strike MX (two Noiseblocker NB-BlackSilent XL1 fans at 1020rpm installed at the bottom front and at the back panel; a default 240mm system fan at 400rpm on the side panel)
• Control and monitoring panel: Ice Hammer IH-4
• Power supply: Thermaltake Toughpower XT 850W (with a default 140 mm fan)
• Monitor: 24-inch BenQ FP241W (widescreen LCD monitor, 1920x1200@60Hz)

To minimize the platform’s influence on the performance of the tested graphics cards I overclocked the quad-core CPU to 4GHz at 1.525V voltage:

The system memory worked at 960MHz with 4-4-4-10 timings (PL=6) and 2.15V voltage.

The overclocked platform’s stability was verified by a half-an-hour run of OCCT version 3.1.0.

The graphics cards from Diamond Multimedia and HIS are copies of the reference cards:

The frequencies of the Diamond Radeon HD 4890 are pre-overclocked. The HIS Radeon HD 4870 X2 has the same frequencies as the reference card.

The tests were carried out under Windows 7 x64 RTM with the following drivers:

• Intel Chipset Drivers 9.1.1.1004 Alpha
• DirectX End-User Runtime March 2009
• Drivers for AMD/ATI graphics cards: Catalyst 9.7 and Catalyst 9.8

When I reinstalled a driver, I also cleaned the system from the remains of the previous version with Driver Sweeper 2.0.5. Once a driver had been installed, I made the following changes in its control panels: image quality changed from “Quality” to “High Quality”, Adaptive Antialiasing set to “Quality”, vertical synchronization set to “Always Off”. There were no other changes.

The graphics cards were tested in two resolutions: 1280x1024 and widescreen 1920x1200. The tests were performed in two image quality modes: “High Quality” without any image quality enhancements and “HQ+AF16x+AA4/8x” with 16x anisotropic filtering and 4x full screen antialiasing (or 8x FSAA if the average frame rate was high enough for comfortable gaming experience). I turned on full-screen antialiasing and anisotropic filtering in the menu of each game. If the game didn’t provide such options, I enabled FSAA and AF from the Catalyst control panel.

The games were installed anew under the new OS and complemented with the latest patches available at the moment. A total of two synthetic benchmarks and 15 games of different genres were used:

• 3DMark 2006 (Direct3D 9/10) – build 1.1.0, default settings (1920x1200, 16x AF and 8x AA);
• 3DMark Vantage (Direct3D 10) – v1.0.2.1, Performance and Extreme profiles (basic tests only);
• World in Conflict (Direct3D 10) – version 1.0.1.0 (b34), “Very High” graphics quality profile, UI texture quality = Compressed; Water reflection size = 512; other settings left at their defaults;
• Enemy Territory: Quake Wars (OpenGL 2.0) – version 1.5, highest graphics quality settings, custom demo “d5” on the Salvage level, Finland;
• Unreal Tournament 3 (Direct3D 9) – version 2.1, highest graphics quality settings (level 5), Motion Blur and Hardware Physics enabled, a flyby of the “DM-ShangriLa” map (two consecutive cycles) using HardwareOC UT3 Bench v1.5.0.0;
• S.T.A.L.K.E.R.: Clear Sky (Direct3D 10.1) – game version 1.5.10, Improved Full DX10 Lighting profile plus 16x anisotropic filtering and other maximum graphics quality settings, custom demo record “s04” (a triple run of the test) on the first game level;
• Crysis (Direct3D 10) – game version 1.2.1, “Very High” settings profile, a double run of the integrated graphics card benchmark;
• Far Cry 2 (Direct3D 10) – version 1.03, Ultra High settings profile, a double run of the Ranch Small test from Far Cry 2 Benchmark Tool (v1.0.0.1);
• Company of Heroes: Tales of Valor (Direct3D 10) – version 2.500, maximum graphics quality settings, a double run of the integrated benchmark (the game was launched with the –novsync key);
• Left 4 Dead (Direct3D 9) – version 1.0.0.5, maximum quality, d3 demo (two runs) on Level 3 called “No Mercy”, first scene called “The Seven”;
• Lost Planet: Colonies (Direct3D 10) – version 1.0, Maximum Quality settings, DX10 HDR Rendering, integrated benchmark including two scenes;
• Cryostasis: Sleep of Reason (Direct3D 10.1) – version 1.0.2, shader model 4.0, high image quality settings, “Caustic” enabled, tested in custom demo “d3” at the “11. Fear” level.
• Call of Duty 5: World at War (Direct3D 9) – version 1.3.1080, graphics and textures are set at “Extra” level, d2 demo on “Dome” level.
• Stormrise (Direct3D 10.1) – version 1.0.0.0, maximum effects and shadows quality, Ambient Occlusion disabled, two runs of the “$mn_sp05” mission demo scene.
• Tom Clancy’s H.A.W.X. (Direct3D 10.1) – version 1.02, maximum graphics quality settings, HDR, DOF and Ambient Occlusion enabled, two runs of the built-in benchmark.
• Call of Juarez: Bound in Blood – version 1.0.0.0, maximum graphics quality settings, Shadow map size = 1024, demo record at the beginning of the Miners Massacre level.
• Resident Evil 5 bench (Direct3D 10.1) – maximum quality settings without motion blur, the result is the average AVG for the four tests; the result of the third test is shown individually.

Unfortunately, I could not launch The Chronicles of Riddick: Assault on Dark Athena under Windows 7 x64 RTM although I tried two game distributions and installed the version 1.01 patch. Thus, I could not benchmark the two cards and two driver versions in that game.

If the game allows recording the bottom frame rate, the latter is also added to the charts. I performed two cycles of tests in each benchmark or game and wrote down the best of the two results if they differed less than 1%. If the difference was bigger, I performed another run of the test to check out the correctness of the result.