Half-Life 2

The benchmark included with the Counter-Strike: Source Beta packs a small level with plenty of pixel shaders and effects to stress your graphics card in an attempt to gauge whether or not your system will be able to run the game. The benchmark does not model actual Half-Life 2 levels, bots or game-play and won't be able to tell you exactly how well Half-Life 2 will run on your computer. However, it is a good tool to measure your graphics card's ability to handle the graphical effects that will most likely be present in the game.

The Source engine stress test is accessed from the Counter-Strike: Source Beta in-game menu. All of the in-game graphical settings were set to high. The "Reflections" drop down menu was set to "Reflect All." Anti-Aliasing and Anisotropic Filtering were controlled through Windows and not the game itself. The stress test runs through the level as fast as it can and spits out the average FPS much like a time-demo. For each test, the stress test was run three times and the results were averaged.

Half-Life 2 Stress Test No AA / No AF
(Show All Graphs) (Collapse Graphs)

1024 x 768

1280 x 1024

1600 x 1200

ATI Radeon X800 Pro

XFX GeForce 6800 GT

PowerColor Radeon 9800 XT

Leadtek WinFast A380 Ultra

88.08

91.10

85.03

87.35

0

(FPS)

111

 

 

1024 x 768

1280 x 1024

1600 x 1200

ATI Radeon X800 Pro

XFX GeForce 6800 GT

PowerColor Radeon 9800 XT

Leadtek WinFast A380 Ultra

77.54

80.33

59.78

56.73

0

(FPS)

111

 

 

1024 x 768

1280 x 1024

1600 x 1200

ATI Radeon X800 Pro

XFX GeForce 6800 GT

PowerColor Radeon 9800 XT

Leadtek WinFast A380 Ultra

67.44

70.07

46.53

43.81

0

(FPS)

111

 

 

Thankfully Doom 3 isn't representative of the X800 Pro's overall performance. Here we see that while the X800 Pro falls slightly behind the 6800 GT with no AA or AF, it does pull ahead when 4x AA and 8x AF are enabled. Both cards look to be very playable at 1600 x 1200 with 4x AA and 8x AF. Let's hope that the Stress Test does a good job of replicating HL2's actual performance.