Always the frustrating test. Indeed, you get 190fps in Quake III, and you're happy if 3DMark sometimes gives you 65. I don't know how they make their benchmarks, but they never even get in sight of Quake III's results.
The full chart of results is as follows :
Game Scores (fps):
| Scene | Low Detail | High Detail |
|---|---|---|
| Car Chase | 103.8 | 39.8 |
| Dragothic | 102.1 | 56.8 |
| Lobby | 106.5 | 50.3 |
| Nature | 33.9 | |
Technical Details :
| Fill Rate | Single Texturing | Multi-Texturing |
|---|---|---|
| MTexels/s | 635,2 | 1319,2 |
| High-Polygon Count | 1 Light | 8 Lights |
|---|---|---|
| MTriangles/s | 15.7 | 3.1 |
| Bump Mapping | Environment | DOT3 |
|---|---|---|
| FPS | 111.6 | 106.2 |
| Vertex Shader Speed | 52.2 FPS |
| Pixel Shader Speed | 77.0 FPS |
| Point Sprites Speed | 14.7 MSprites/s |
The end result is a 3DMark score of 6737 points. That is a full 310 points (4.6%) less than the scores given by Tom's Hardware, but these guys are professionals, so their platform must be better than mine.
What is important is that Sharok's new score is 1314 points higher than with the T-Bird (5423 points previously). That means a 24.3% increase in performance, somewhat consistant with the Quake III performance increase of 24%.