Detonator 52.14 Test & 52.10 Re-Test
October 8, 2003 / by Leonidas / page 4 of 7
Game Performance
So far about the "optimizations", now let's focus on the pure performance of the new drivers. Through their different degrees of "optimization", the detonators 45.23, 51.75 and 52.10/14 are hardly comparable with each other, indeed only 52.10 and 52.14 are. In spite of that we didn't want to leave this driver's review without any benchmarks.
The test machine was a Pentium4 2.53GHz (FSB533) on a MSI 845PE Max mainboard with i845PE chipset and 512MB PC3200 DDR-RAM running WindowsXP with Service Pack 1 and DirectX 9.0b installed. Video card was a MSI GeForceFX 5900 with 128MB RAM. All benchmarks were run at a resolution of 1024x768 with maximum settings and 8x anisotropic filtering enabled. Anisotropic filtering was activated using the Control Panel since this is the only way to force all test applications to use anisotropic filtering and it should also be the most common aproach of the users of nVidia graphics adapters:
45.23 8xAF (CP) |
51.75 8xAF (CP) |
52.10 8xAF (CP) |
52.14 8xAF (CP) |
|
AquaNox 2 | 58.3 fps | 76.6 fps | 78.2 fps | 77.4 fps |
Comanche 4 | 41.4 fps | 43.0 fps | 43.1 fps | 42.9 fps |
Dungeon Siege | 66.8 fps | 68.3 fps | 69.1 fps | 68.6 fps |
Elite Force II | 54.5 fps | 55.8 fps | 55.8 fps | 55.8 fps |
Enemy Territory | 57.0 fps | 57.0 fps | 56.9 fps | 55.9 fps |
Max Payne | 55.1 fps | 56.3 fps | 56.9 fps | 58.7 fps |
Serious Sam: TSE | 71.8 fps | 66.6 fps | 67.0 fps | 67.4 fps |
Unreal Tournament 2003 | 44.3 fps | 45.6 fps | 45.7 fps | 45.3 fps |
... and additionally the scores using 8x anisotropic filtering with 4x Anti-Aliasing, also activated by Control Panel:
45.23 8xAF/4xAA (CP) |
51.75 8xAF/4xAA (CP) |
52.10 8xAF/4xAA (CP) |
52.14 8xAF/4xAA (CP) |
|
AquaNox 2 | 38.8 fps | 56.9 fps | 57.6 fps | 57.3 fps |
Comanche 4 | 39.0 fps | 41.1 fps | 41.3 fps | 40.5 fps |
Dungeon Siege | 54.9 fps | 51.1 fps | 58.5 fps | 58.2 fps |
Elite Force II | 43.3 fps | 44.5 fps | 44.5 fps | 44.5 fps |
Enemy Territory | 50.8 fps | 51.2 fps | 51.1 fps | 50.6 fps |
Max Payne | 50.4 fps | 51.2 fps | 52.4 fps | 53.1 fps |
Serious Sam: TSE | 68.9 fps | 62.2 fps | 62.9 fps | 62.4 fps |
Unreal Tournament 2003 | 33.3 fps | 34.2 fps | 35.0 fps | 34.5 fps |
The about +7% (8x anisotropic filtering) and +8% (8x anisotropic filtering with 4x Anti-Aliasing) higher scores of Detonator 52.10 versus 45.23 (or +2.5% and +4% without AquaNox 2 and Serious Sam: TSE, which deliver quite extreme scores) seem to be a direct result of the dicovered "optimizations". Whether these scores also reflect real driver improvements (which do NOT decrease image quality) cannot be clarified because there is no way of comparing drivers 45.23 and 52.10 with identical image quality.
We want to remark additionally that the new driver 52.14 is in general a bit slower than 52.10 while delivering identical image quality. Although we are talking about a difference of -0.2% (8x anisotropic filtering) and -0.6% (8x anisotropic filtering with 4x Anti-Aliasing) which is actually under the treshold of measurement accurancy.