A PARALLEL ARCHITECTURE FOR RAY TRACING by

Ray-tracing techniques for image rendering have produced some of the most realistic images to date. They are also slow because the process of tracing a ray is computationally intensive and because there are many rays to be traced. However. since computations for each pixel are independent, ray tracing is amenable t o parallel processing uslng image space subdivision. The processor per subdivision approach is unattractive because r d e~ther the large memory requirements per processor or the high communication oierhead of a shared memory, We present an architecture that addresses these drawbacks using broadcasting The architecture is based on. (a) an interconnection of multiple ray tracing engines working in parallel; (b) 3 disjoint data sets resulting from the use of a modified hierarchical data structure-based ray tracing algorithm: and (c) 3 broadcast processors each with its own memory module and broadcast bus. Simulation results show substantial rendering time improvements over mini-computer timings.