4 Conclusions

In this work we implemented a parallel Radiosity rendering algorithm on a loosely-coupled cluster of PCs. The initial naive algorithm did not perform well, and in fact the computation time increased with the number of machines. We then applied a handful of tunings and optimization techniques which eventually made the algorithm execute eeciently, achieving speedups close to optimal. Previous related works such as 8] use a diierent approach in which the given scene is divided into sub-scenes, each assigned to a diierent processor. The communication pattern thus reeects the energy casting from one sub-scene to another, and seems suitable for implementation in a message passing parallel system. As described throughout the paper, our approach seems suitable for parallelization over a distributed shared memory environment such as millipede. This project was carried on the millipede system which provides an easy to use environment for parallel programming on top of distributed clusters of PCs. In fact, the millipede concept was designed to promote this precise modus operandi (in the process of parallel programming): a rst easy phase of porting an existing sequential code to the parallel domain, and then, in a follow-up phase, a step-by-step gradual optimization towards eeciency and improved speedups. Delta delta values for naive scheduler running a scene on 5 machines Figure 7: The scheduler sleeps for milliseconds between successive iterations. The optimal for the scene benchmark turns out to be approximately 100 milliseconds. Execution time using ve machines is very high when values are much larger than the optimal one.