DMCA
Binned SAH Kd-Tree Construction on a GPU (2010)
| Citations: | 3 - 0 self |
BibTeX
@MISC{Danilewski10binnedsah,
author = {Piotr Danilewski and Stefan Popov},
title = {Binned SAH Kd-Tree Construction on a GPU},
year = {2010}
}
Share
 |
 |
 |
 |
||
OpenURL
Abstract
In ray tracing, kd-trees are often regarded as the best acceleration structures in the majority of cases. However, due to their large construction times they have been problematic for dynamic scenes. In this work, we try to overcome this obstacle by building the kd-tree in parallel on many cores of a GPU. A new algorithm ensures close to optimal parallelism during every stage of the build process. The approach uses the SAH and samples the cost function at discrete (bin) locations. This approach constructs kd-trees faster than any other known GPU implementation, while maintaining competing quality compared to serial high-quality CPU builders. Further tests have shown that our scalability with respect to the number of cores is better than of other available GPU and CPU implementations. 1
