Benchmarking for ray tracing performance measurement

Abstract

During the past thirty years, researchers have greatly sped up the naïve ray tracing algorithm. However, the methods used to measure and compare their diverse algorithms have remained ad-hoc. This makes it difficult to gauge the effectiveness of one author's approach versus that of another. Clarity, transparency and the reproducibility of experiments are foundational to proper science, therefore these discrepancies must be addressed. This paper will discuss a variety of sources from the literature to show the flaws in the current research paradigm. It will then propose a new method based on a benchmarking approach. Finally, the results of experimentation with the approach will be discussed and a conclusion will be drawn as to its applicability.