Benchmarks - What We Use And Why
3DMark 2005 And 2003
3DMark is called the "Gamer's Benchmark," and THG has been using its different versions for years. The mark is set using a reference system; the higher the score, the better the performance of that system versus the reference one. With each new version of 3DMark, scores are scaled down so they do not become astronomically high. These third-party benchmarks are widely used across the industry.
Unreal Tournament 2004
Many games are based on one of the Unreal Engines. For this reason, we feel it is necessary to keep UT2004 in the lineup. We use a custom time demo for UT2004, set up in assault mode on the Fallen City map. We do not use bots, and the graphics details are of course kept at their highest level.
Far Cry 1.3 (32-bit) - Cooler01 Demo
The Crytek engine is just one of many, but Far Cry is a good game to tax the graphics card with textures, shading and lighting. We turned on the flashlight in our custom time demo, Cooler01, to make sure it got a thorough workout.
Doom 3 - Demo1
The next game engine to come onto the horizon was the Doom 3 engine, which will put cards through an ordeal with its complex lighting. We use the Doom 3 default time demo, "demo1," for each of our tests. First, the graphics card driver is set to "application controlled." Then we disable FSAA and the high-quality setting in the game for the first run. When set at high quality, anisotropic filtering is automatically enabled and set to eight-sample mode. For the second run, we turn FSAA to 4X to really tax the GPU.
Half-Life 2 - MW_Coast1 Demo
Last but not least, we have Valve's Source Engine, which drives Half-Life 2. HL2 is a game that offloads some of the work to the CPU, so it is a good test for the latest graphics cards as well as overall system performance. And as a bonus, HL2 offers some serious "eye candy" for the gamer.