Why does SBVH outperform KD-tree on parallel platforms?
Alfonso Breglia, Amedeo Capozzoli, Claudio Curcio, Angelo Liseno · 2016
We spot on the performance of two acceleration data structures for electromagnetic ray tracing purposes on GPU using the CUDA programming language, namely the KD-tree and the SBVH. Our implementations have been based on the approach made available by NVIDIA which takes into account for the programming optimizations made possible by the latest version of CUDA and for the latest NVIDIA GPU architectures.