By Lars Weinand / Thomas Pabst (Tom´s Hardware Guide)

Highlight: Dual Vertex Shaders

If you remember the GeForce3's "nfiniteFX"engine, you know what the GeForce4 TI's new "nfinite FX II" engine is. This is the name NVIDIA gave to íts first programmable pixel and vertex shader, which was introduced in the GeForce3. (You can find detailed information on Vertex Shaders at Tom´s Hardware Guide).

Probably the single most important change in the GeForce4 design is the inclusion of a second vertex shader, a concept introduced in the Xbox's X-GPU. NV25 builds on this concept, using a further refined design.

Obviously, two vertex shaders will be able to process more vertices than a single shader could. Additionally, "nfiniteFX II" can process several vertices or instructions in parallel ("multi-threading"). This functionality is native to the GF4 chip in hardware and is therefore transparent to any applications or APIs, meaning that applications do not need to be re-programmed to take advantage of this feature. The only requirement for effective parallel processing is that the individual shaders work on different vertices. The NV25 itself takes care of this regulation.

To sum it up, the GeForce4 TI4600 can process roughly three times as many vertices as the GeForce3 TI500, since it has two vertex shaders which are more refined and run at higher speeds.

Copyright: 06.02.2002 -   RIVA Station 2002 - Lars Weinand
No Copy without Permission!

URL of this Article: www.rivastation.com/geforce4_e.htm - If you want to link to it, please use this URL! :-)