Half Life 2
Here we'll use Half Life 2 to test the DirectX9 rendering performance of the graphics boards in this title. We are using our own internal benchmark from a level that has a lot of shader usage in order to maximise the utilisation of the graphics board, rather than the CPU.
| Half Life 2 (FPS) | 640x480 | 800x600 | 1024x768 | 1280x960 | 1600x1200 |
| S18 Nitro | 36.6 | 23.3 | 14.0 | 10.7 | 6.4 |
| X600 PRO | 83.5 | 56.1 | 33.8 | 24.9 | 14.5 |
| 6600 | 86.4 | 63.2 | 41.5 | 32.5 | 20.6 |
| 6200 | 56.6 | 38.9 | 24.3 | 18.6 | 11.6 |
| S18 % Faster Than: | 640x480 | 800x600 | 1024x768 | 1280x960 | 1600x1200 |
| X600 PRO | -56.2% | -58.5% | -58.6% | -56.9% | -55.7% |
| 6600 | -57.7% | -63.2% | -66.3% | -67.0% | -68.8% |
| 6200 | -35.5% | -40.2% | -42.4% | -42.6% | -44.8% |
This particular Half Life 2 test uses plenty of DirectX9 shaders and we can see again that the performance of the S18 is bringing up the rear, even more so that in the other cases with as much as nearly a 70% deficit to the 6600.
| Half Life 2, 2x FSAA & 4x AF (FPS) | 640x480 | 800x600 | 1024x768 | 1280x960 | 1600x1200 |
| S18 Nitro | 23.4 | 14.8 | 8.8 | 6.3 | 3.9 |
| X600 PRO | 60.3 | 39.6 | 22.6 | 17.3 | 8.8 |
| 6600 | 69.8 | 50.9 | 33.3 | 25.6 | 15.7 |
| 6200 | 45.7 | 31.6 | 19.6 | 14.8 | 9.2 |
| S18 % Faster Than: | 640x480 | 800x600 | 1024x768 | 1280x960 | 1600x1200 |
| X600 PRO | -61.2% | -62.7% | -61.1% | -63.6% | -55.9% |
| 6600 | -66.5% | -71.0% | -73.5% | -75.4% | -75.4% |
| 6200 | -48.8% | -53.2% | -55.1% | -57.5% | -57.8% |
The performance gap widens a little further when 2x FSAA and 4x AF are enabled - SuperSampling results in everything getting sampled a multiple of times by the factor of the FSAA level, so for each pixel that is shaded the entire thing is being calculated twice on the GammaChrome; MultiSampling only has to sample the pixels at triangle edges a multiple times which results in less work needing to be done overall.
| S18 Nitro (FPS) | 640x480 | 800x600 | 1024x768 | 1280x960 | 1600x1200 |
| Normal | 36.6 | 23.3 | 14.0 | 10.7 | 6.4 |
| 4x AF | 36.2 | 22.9 | 13.6 | 10.5 | 6.3 |
| 2x FSAA | 23.5 | 14.9 | 9.0 | 6.5 | 4.0 |
| 2x FSAA & 4x AF | 23.4 | 14.8 | 8.8 | 6.3 | 3.9 |
| % Diff from Normal | 640x480 | 800x600 | 1024x768 | 1280x960 | 1600x1200 |
| 4x AF | -0.9% | -1.5% | -3.0% | -1.6% | -2.2% |
| 2x FSAA | -35.6% | -35.8% | -35.6% | -39.2% | -37.9% |
| 2x FSAA & 4x AF | -36.0% | -36.5% | -37.0% | -41.1% | -39.6% |
As we've seen in all the game tests, bar UT2004, enabling 4x AF in this test results in a very small performance drop, whilst FSAA a much larger one. However, the performances with the very high details settings are very low, expect at low resolution.
