Reservoir: An Alternative Load Balancing Technique for Parallel Ray Tracing

Emulating the real world is the next frontier in computer applications. However, a computing machine cannot accurately imitate reality in a decent amount of time or even apart of it single-handedly. The solution: parallelize computers to solve a specific problem. One such aspect of reality that can be replicated is light. Ray tracing is a rendering technique that accurately depicts light at the cost of time. However, simply parallelizing it is not enough to reduce its running time. There is also the issue of having multiple heterogeneous computers that would divide the task unevenly. Thus, a load balancing technique must be performed to further increase efficiency and make ray tracing suitable for mainstream purposes. The goal of this project is to parallelize ray tracing and compare different implementations to efficiently distribute the load and render images in the smallest possible amount of time given in a network of machines. This project proposed an alternative load balancing technique, Reservoir, to address the issue of having heterogeneous computers within the network and minimizes the communication overhead implicitly imposed by parallel applications.