The feature of the GeForce2 GTS that sets it apart from all other current consumer video cards is its hardware transformation and lighting (T&L) engine. The GeForce2 GTS, clocked at 200MHz, can deliver over 25 million lit polygons to the screen per second. This can be compared to the GeForce 256, which managed a relatively weak 15 million. By loading T&L calculations off the CPU, more CPU power can be dedicated to AI and other game features at the same time as geometric complexity increases.
NVIDIA says that a major new feature of the GeForce2 GTS is its per-pixel shader. Vertex lighting, the most common shading method in modern games, is where a pixel's shading is calculated from the three vertices of the polygon it is located on. This is a quick and decent looking method of shading, however, it does not look so hot when you have large polygons on screen.
NVIDIA's per-pixel shading enables the computer to calculate shading for each pixel on their own through the use of a "normal map." Through the use of a normal map, a GeForce2 GTS can render more accurate shading than vertex lighting. It also enables quick and attractive bump mapping. The problem with per-pixel shading is that a developer has to write software specifically to support it.
Want more information on per-pixel shading? You can read an in-depth look at per-pixel shading inside our GeForce2 GTS Guide.
Of course, the most exciting part about the GeForce2 GTS is its awesome fill rate and texel rate. With four pipelines capable of drawing one pixel per clock with two textures, the card can draw an awesome 800 m/pixels per second and 1600 m/texels per second. With multi-textured games like Quake III Arena, awesome gains in speed can be realized.