Abstract
The design of a computer graphics pipeline is examined in detail using high speed floating point multiply/add units and cross bar switches as building blocks. Various parallel and pipelined arrangements are explored for simulating the individual stages of the pipe. It is shown that a highly versatile architecture, entitled the FLEX ARCHITECTURE based on four processors and a 16 port switch, can be used efficiently to simulate all stages of the graphics pipeline and, for high performance systems, multiple flex units are considered. Estimates of performance are developed using a particular set of components available from Texas Instruments. Designs with performance in excess of one to two million vectors per second are feasible.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call