Conference Main
Course 3

Parallel and Distributed Photo-Realistic Rendering

Parallel processing offers an approach that significantly reduces computation time and exploits combined memory resources. Complex environments must be distributed over the processors' local memories, so careful attention must be paid to the correct management of data and tasks, to ensure an efficient and scalable solution. This course addressed techniques for decreasing the volume of cached data, aiding memory management, and optimizing communication. It included task scheduling and memory management techniques for three photo-realistic graphics algorithms: ray tracing, radiosity, and particle tracing. Their different parallel implementation approaches were compared and contrasted.

Prerequisites
Knowledge of ray tracing, radiosity, particle tracing, geometric simplification, levels of detail, impostors, and related topics. Experience with parallel processing useful, but not assumed.

Topics Covered
Task migration, coherence, load balancing, and communication issues, as well as data decomposition, data redistribution, and various caching mechanisms. Their application to ray tracing, radiosity, and particle tracing was compared and contrasted.

Organizer
Alan Chalmers
University of Bristol

Lecturers
Alan Chalmers
Erik Reinhard
University of Bristol

Main Comments and questions