Testbed and Methods
Every test was performed in a closed system case with the following configuration:
- Mainboard: ASUS P6T Deluxe (Intel X58 Express, LGA1366, BIOS 2101)
- CPU: Intel Core i7-920, 2.67 GHz (Bloomfield, C0, 1.2 V, 4x256 KB L2 cache, 8 MB L3 cache)
- Cooling system: Xigmatek Balder SD1283 (with two Thermalright TR-FDB fans at 1100 RPM)
- Thermal interface: Arctic Cooling MX-2
- System memory: DDR3 3x2 GB Wintec AMPX 3AXH1600C8WS6GT (1600 MHz / 8-8-8-24 / 1.65 V)
- System disk: SSD OCZ Agility EX (SATA-II, 60 GB, SLC, Indilinx, firmware v1.31);
- Hard disk for games and benchmarks: Western Digital VelociRaptor (SATA-II, 300 GB, 10,000 RPM, 16 MB cache, NCQ) in Scythe Quiet Drive 3.5” enclosure
- Archive disk: Western Digital Caviar Green WD10EADS (SATA-II, 1000 GB, 5400 RPM, 32 MB, NCQ)
- System case: Antec Twelve Hundred (front panel: three Noiseblocker NB-Multiframe S-Series MF12-S1 fans at 900 RPM; back panel: two Scythe SlipStream 120 fans at 900 RPM; top panel: the default 200mm fan at 400 RPM)
- Control and monitoring panel: Zalman ZM-MFC2
- Power supply: Zalman ZM1000-HP 1000 W, 140mm fan
- Monitor: 30” Samsung 305T Plus
To avoid being limited by the CPU's performance, the 45nm quad-core CPU was overclocked to 4.0 GHz (at a frequency multiplier of 21, voltage of 1.3725 V, and with Load-Line Calibration enabled) in some games included into this test program.
The computer's system memory was working at 1.53 GHz with 7-7-7-14_1T timings at a voltage of 1.64 V. The rest of the BIOS settings pertaining to CPU and memory overclocking were left at their defaults (“Auto”).
AMD's camp will be represented by a Radeon HD 5870 1024 MB and by a dual-processor Radeon HD 5970 2x1024 MB.
Now let's move on to the software and benchmarking tools used in the tests. I started the test session on the 23rd of March and carried it out under the Windows 7 Ultimate x64 operating system with all critical updates installed. I used the following drivers:
- Intel Chipset Drivers 18.104.22.1685 WHQL for the mainboard chipset
- DirectX End-User Runtimes from February 2010
- Catalyst 10.3 drivers for graphics cards with AMD's GPUs
- GeForce/ION Driver 197.17 beta for the GeForce GTX 480 and GeForce/ION Driver 197.25 beta for the other Nvidia-based graphics cards
- Nvidia PhysX System Software 9.10.0129 physics effects acceleration driver
The graphics cards were tested at two resolutions: 1920x1080 and 2560x1600. I believe that it wouldn't make sense to test the premium-class products at lower resolutions as it wouldn't bring any valuable results, the cards being often limited by the speed of the platform.
The tests were performed in two image quality modes: “High Quality + AF16x” (maximum textures with 16x anisotropic filtering) and “High Quality + AF16x + AA 4(8)x” with 16x anisotropic filtering and full-screen antialiasing (MSAA) of 4x or 8x level (if the average frame rate remained high enough for comfortable play). I enabled anisotropic filtering and full-screen antialiasing in the games' settings or configuration files. If such options were missing, I enabled these settings in the Control Panels of the Catalyst and GeForce drivers. Vertical sync was always off in the drivers' control panels.
All games were updated with the latest patches available at the time of tests. So, the complete list of test applications includes two popular semi-synthetic benchmarking suites, one technical demo and 21 games, including the latest titles. Here is the complete list of tests with the settings (the games are sorted in the order of their release):
- 3DMark 2006 (DirectX 9/10) - build 1.2.0, default settings and 2560x1600 with 16x AF and 8x FSAA;
- 3DMark Vantage (DirectX 10) - v22.214.171.124, Performance and Extreme profiles (basic tests only);
- Unigine Heaven Demo (DirectX 11) - version 2.0, maximum graphics quality settings, “extreme” tessellation;
- World in Conflict (DirectX 10) - version 126.96.36.199 (b34), “Very High” graphics quality settings, Water Reflection Clouds turned on, built-in benchmark;
- Crysis (DirectX 10) - game version 1.2.1, “Very High” settings profile, two runs of “Assault harbor” demo from Crysis Benchmark Tool version 188.8.131.52;
- Unreal Tournament 3 (DirectX 9) - version 2.1, highest graphics quality settings (level 5), Motion Blur and Hardware Physics enabled, a flyby of the “vCTF-Corruption” map (two consecutive cycles) using HardwareOC UT3 Bench v184.108.40.206;
- Lost Planet Extreme Condition: Colonies Edition (DirectX 10) - version 1.0, Maximum Quality settings, DX10 HDR Rendering, integrated benchmark including two scenes, but the results are provided from the first scene only (ARENA 1);
- Far Cry 2 (DirectX 10) - version 1.03, “Ultra High” settings profile, two runs of the Ranch Small test from Far Cry 2 Benchmark Tool (v220.127.116.11);
- Call of Duty 5: World at War (DirectX 9) - version 1.6, graphics and textures are set at “Extra” level, Breach demo from the same-name level;
- BattleForge: Lost Souls (DirectX 11) - version 1.2 (19.03.2010), maximum image quality settings, shadows enabled, SSAO technology enabled, two runs of the built-in benchmark;
- Stormrise (DirectX 10.1) - version 18.104.22.168, maximum effects and shadows quality, Ambient Occlusion disabled, two runs of the “$mn_sp05” mission demo scene;
- Tom Clancy's H.A.W.X. (DirectX 10) - version 1.03, maximum graphics quality settings; HDR, DOF and Ambient Occlusion enabled, two runs of the built-in benchmark;
- Call of Juarez: Bound in Blood (DirectX 10.1) - version 22.214.171.124, maximum graphics quality settings, Shadow map size = 1024, 110-second demo in the beginning of “Miners Massacre” level;
- Wolfenstein MP (OpenGL 2.0) - version 1.3, maximum graphics quality settings, custom d2 demo recording on “Manor” level;
- Batman: Arkham Asylum (Direct3D 9) - version 1.1, maximum level of detail, maximum physics effects, a double run of the built-in benchmark;
- Resident Evil 5 (DirectX 10.1) - version 1.0, variable benchmark with maximum graphics quality settings without motion blur, we take the average result of the heaviest third scene for analysis;
- S.T.A.L.K.E.R.: Call of Pripyat (DirectX 11) - version 1.6.02, Enhanced Dynamic DX11 Lighting profile with all parameters manually set at their maximums, we use our custom cop03 demo on the Backwater map;
- Borderlands (DirectX 9) - version 1.2.1, “timedemo1_p” with maximum image quality settings;
- Left 4 Dead 2 (DirectX 9) - version 126.96.36.199, maximum graphics quality settings, d333 demo (two runs) on “Swamp Fever” map of the “Swamp” level;
- Colin McRae: DiRT 2 (DirectX 9/11) - version 1.1, built-in benchmark including two laps of the “London” track with maximum graphics quality settings;
- Wings Of Prey (DirectX 9) - version 188.8.131.52, “Ultra High” graphics quality and other maximum settings, two-minute demo on “Escort” level from the “Battle of Britain” campaign;
- Warhammer 40 000: Dawn of War II - Chaos Rising (DirectX 10.1) - version 184.108.40.20679, Ultra graphics quality settings, three or four runs of the built-in benchmark;
- Metro 2033 (DirectX 10/11) - version 1.0, maximum graphics quality settings, we use a scripted 160-second scene on the Chaser map, a double sequential run;
- Just Cause 2 (DirectX 11) - version 220.127.116.11, maximum quality settings, Background Blur and GPU Water Simulation off, a double sequential ruin of the Dark Tower demo.
If the game allowed recording the bottom fps readings, they were also added to the charts. I ran each game test or benchmark twice and took the best result for the diagrams, but only if the difference between them didn't exceed 1%. If it did exceed 1%, I ran the tests at least one more time to achieve repeatability of results.