Exploiting Spatial, Ray, and Frame Coherency for Efficient Parallel Volume Rendering

Coherency (data locality) is one of the most important factors that influences the performance of distributed ray tracing systems, especially when object dataflow approach is employed. The enormous cost associated with remote fetches must be reduced to improve the efficiency of the parallel renderer. Objects once fetched should be maximally utilized before replacing them with other objects. In this paper, we report on the results obtained from the implementation of two coherent parallel rendering algorithms on distributed memory architectures. The algorithms are tested with several experimental datasets, and the timings delineate the coherent nature of our algorithms. Several network issues like latency hiding, communication overheads, load balancing, scalability, and limited channel bandwidth are considered during the design process. Our algorithms optimize communication between nodes, eliminate the need for any explicit synchronization, and support latency hiding in an efficient mann..