FSAA Performance
Finally, we'll take a look at the Radeon X800's rendering performance with FSAA under a few different scenarios.
| RightMark - Lighting (Phong) - PS2.0, 1024x768 FSAA (FPS) | No AA | 2X AA | 4X AA | 6X AA |
| X800 XT PE | 277.9 | 269.4 | 265.7 | 265.0 |
| X800 PRO | 190.4 | 185.7 | 182.9 | 182.1 |
| % Drop from 1X | 2X AA | 4X AA | 6X AA |
| X800 XT PE | -3.1% | -4.4% | -4.6% |
| X800 PRO | -2.5% | -3.9% | -4.4% |
| % Drop from Previous | 2X AA | 4X AA | 6X AA |
| X800 XT PE | -3.1% | -1.4% | -0.3% |
| X800 PRO | -2.5% | -1.5% | -0.4% |
Testing the FSAA performance under a heavy pixel shader test, we can again see that due to the length of instructions, hence the time spent in the shaders generating the final pixel result, we see that the bandwidth penalty of FSAA is nigh on negated.
| 3DMark03 GT2, 1600x1200 FSAA (FPS) | No AA | 2X AA | 4X AA | 6X AA |
| X800 XT PE | 49.8 | 34.5 | 19.9 | 13.7 |
| X800 PRO | 35.7 | 22.6 | 14.5 | 9.3 |
| % Drop from 1X | 2X AA | 4X AA | 6X AA |
| X800 XT PE | -30.7% | -60.0% | -72.5% |
| X800 PRO | -36.7% | -59.4% | -73.9% |
| % Drop from Previous | 2X AA | 4X AA | 6X AA |
| X800 XT PE | -30.7% | -42.3% | -31.2% |
| X800 PRO | -36.7% | -35.8% | -35.9% |
In the more varied environment of 3DMark03’s we can see that enabling FSAA is incurring a much greater performance penalty, due to the heavy bandwidth demands.
