Catalyst 3.0 under the microscope: Are all good things really 3.0?

Catalyst 3.0 under the microscope: Are all good things really 3.0?


'Industry's most stable 3D acceleration software' so it says on the product page of ATi's recently released Catalyst 3.0 driver package.

The same has already been proud once announced, namely with the release of the original driver software called Catalyst. A short time later, however, this addition was removed from the website.

Whether the repeated full-bodied announcement by ATi is true, but above all, to what extent there is actually a performance and function difference between the Catalyst 2.5, which is still based on DirectX8 Drivers and the new Catalyst 3.0 drivers, which already require DirectX9, we want to find out in this short driver comparison.

Test system and benchmarks

The test system consisted of a combination of SiS745 chipset and AthlonXP 1800+ with 512MB DDR-RAM under WindowsXP.

The onboard sound of the SiS chipset with drivers 7012.106 is mandatory for sound generation, because what use are the most beautiful gains in raw mode when in-game Nothing happens due to practical influences.

In order to clarify the differences, we use a Radeon9500 with 64MB memory. On the one hand, this model shouldn't be slowed down too much by our AthlonXP.On the other hand, due to its DirectX9 design, it could possibly benefit more from this interface than, for example, an older Radeon9000, even if DirectX8 applications continue to run on the separately available, but apparently newly compiled d3d8.dll> CPU:

  • AMD AthlonXP 1800+
  • Motherboard:
    • Elitegroup K7S6A with SiS745 chipset (SiS AGP driver 1.11)
  • RAM:
    • 2 * 256MB Nanya/Apacer PC2100 DDR-RAM CL2 with the sharpest settings
  • Graphics card:
    • Radeon9500 64MB with the following driver versions
    • 6200 (offz. Catalyst 2.4, still tested under DirectX8.1 !!)
    • 6218 (offz. Catalyst 2.5)
    • 6255 (offz. Catalyst 3.0)
  • Microsoft DirectX 9 final
  • For the purpose of a driver comparison we have limited our usual wide range of benchmarks a little. Since this is only a single product, only the differences within a given environment are of interest, so that further benchmarks would only have inflated the comparison unnecessarily.

    Synthetic benchmarks
    • 3DMark 2001 SE v3.30
    • Templemark D3D v1.06
    • Aquamark D3D v2.3
    Game benchmarks
    • Comanche 4 Demo D3D
    • Dungeon Siege D3D
    • Aquanox D3D v1.07
    • Jedi Knight II OGL v1.04
    • Max Payne D3D v1.02
    • Serious Sam SE OGL v1. 07
    • Unreal Tournament 2003 Demo

    Unless otherwise noted, all benchmarks ran in 32-bit color depth and if possible also 32-bit textures. In general, the highest possible detailed settings were chosen to really challenge the test subjects.

    More details on the individual benchmarks and their settings can be found in our other graphics card reviews .

    Upthe next page: Synthetic Benchmarks