A Parallel Finite Oct-tree for Multi-threaded Insert, Delete, and Search Operations

We present an object-oriented approach to distribute simulation data (like e.g., geometrical information, meshing information, distributed quantities) used for the simulation of semiconductor fabrication processes such as Monte-Carlo ion implantation [1], etching [2, 3), diffusion [4], and oxidation [5] on a heterogenous cluster of workstations. The approach is based on a finite oct-tree where the data are spread over the network. On SYMMETRIC MULTI PROCESSING (SMP) machines several elements can be handled in parallel (multi-threaded) to speed up insertion, searching, and deletion.