A parallel processor architecture for graphics arithmetic operations

Interactive 3D graphics applications require significant arithmetic processing to handle complex models, particularly if realistic rendering techniques are used. Current semiconductor technology cannot provide the necessary performance without some form of multi-processing.This paper describes a graphics processor architecture which can be configured with an arbitrary number of identical processors operating in parallel. Each of the parallel processors can be programmed identically as if it were a single processor system, substantially simplifying software development and allowing complex rendering functions to take advantage of the multiple processors. The architecture described is able to achieve extremely high performance while allowing the price/performance of the system to be optimized for a given application.Techniques are described for handling graphics command distribution, sequencing of commands which must be processed in order, parallel processing of graphics primitive picking, and handling inquiry (read-back) commands.