A fast SAH-based construction of Octree

Abstract

Recent GPU ray tracers can already achieve performance competitive to that of their CPU counterparts. Nevertheless, these systems can not yet fully exploit the capabilities of modern GPUs and can only handle medium-sized, static scenes. We present an octree construction algorithm for the GPU that achieves real-time performance by heavily exploiting the hardware, which has been observed to give superior performance in ray tracing compared to other acceleration structures. We use streaming construction with the surface area heuristic (SAH) that significantly increase the coherence of memory accesses during construction of the octree.