Summary: A Case Study of Load Distribution in Parallel
View Frustum Culling and Collision Detection
and Per StenstrĻom2
ABB Robotics, Drakegatan 6, SE-412 50 GĻoteborg, Sweden
Department of Computer Engineering Chalmers University of Technology SE-412
96, GĻoteborg, Sweden
Abstract. When parallelizing hierarchical view frustum culling and col-
lision detection, the low computation cost per node and the fact that the
traversal path through the tree structure is not known `a priori make the
classical load-balance versus communication tradeoff very challenging.
In this paper, a comparative performance evaluation of a number of load
distribution strategies is conducted. We show that several strategies suf-
fer from a too high an orchestration overhead to provide any meaningful
speedup. However, by applying some straightforward tricks to get rid of
most of the locking needed, it is possible to achieve interesting speedups.