Quake 3 has been used for benchmarking video cards for some time now.  It has become one of the standards for real time tests but it is starting to show its age. I decided some time ago to switch to Return to Castle Wolfenstein to test real time performance instead of Quake 3 because it seams to be able to push the video card a little harder then Quake 3 can and since RTCW is based on the Quake 3 engine I can still get the results that I would be looking at Quake 3 for. In this test though, it doesn't seem like AGP aperture affects the performance at all.

The makers of Unreal where nice enough to package this game with a benchmark. I really wish all game manufacturers would do this. I think it would help them sell games because sites like mine would be using them as benchmarks. Hey free advertising but I guess companies don't see it that way or there afraid of what people might think if there benchmark shows poor performance, who knows, but I am still glad that UT2003 has came with a benchmark. According to my testing in this test it seems that at low resolutions the Flyby improves performance slightly when the AGP aperture is raised but when the Flyby is running at higher resolutions it seems to loose performance as the AGP aperture gets over 128Mb. Now Bot Match seems the exact opposite. When playing the games at lower resolutions the performance drops slightly when the AGP aperture is raised and the performance increases when higher resolutions have a larger AGP aperture.

Comanche 4 uses some nice special effects and I was hoping it would bend the charts a little in these tests but unfortunately it seems that Comanche 4 doesn't use enough textures to effect the AGP aperture size at all.