This work presents a mesh generation strategy for large scale distributed parallel computations. The mesh generation algorithm based on an octree decom- position of arbitrarily complex geometries is presented and the format used to store the mesh data is detailed. The resulting mesh data encoding the complete tree information is very compact and mostly requires only 16 bytes per element. The format is well suited for parallel processing, especially parallel IO to avoid serialization all over the simulation, even in initialization. It is highly portable by the usage of simple 8 byte signed integer numbers only. As input to the mesh generator the widely used STL data format is chosen for geometry representations. The described software Seeder is available under an open source license.
CITATION STYLE
Harlacher, D. F., Hasert, M., Klimach, H., Zimny, S., & Roller, S. (2011). Tree Based Voxelization of STL Data. In High Performance Computing on Vector Systems 2011 (pp. 81–92). Springer Berlin Heidelberg. https://doi.org/10.1007/978-3-642-22244-3_6
Mendeley helps you to discover research relevant for your work.