Data structures for ray tracing on Cell architecture

Today, streaming multiple core architectures are on the rise. This brings an option to use these architectures for ray tracing as these algorithms can heavily benefit from multiple available cores. We show an implementation of two different acceleration structures on the Cell Broadband Engine Architecture, which belongs to the aforementioned group. We show the limitations of Cell and how some of them may be overcome by software caches, branch hinting and branchless code.