Comparing Two Continuous Collision Detection Algorithms on CPU and GPUs

Fast filtering algorithms can highly improve the narrow-stage performance of continuous collision detection (CCD). The algorithm by Tang et al. is a fast solution to this problem, which casts it into a geometric constrain. We show that, after slight modification, this earlier work is interestingly equivalent to the recent result by Zhang and Liu that is directly derived from algebraic theorems. We compare the computational complexity and performance for the two algorithms using cloth simulation on CPU and GPUs.