Parallel Implementation of Mesh Simplification on a Beowulf Cluster

The parallel implementation of a novel mesh simplification method is introduced detailedly in this paper, which is based on a Beowulf cluster system. Taking full advantage of the distributed memory and high performance network, we can simplify out-of-core models quickly and avoid thrashing the virtual memory system. In addition, the file I/O and load balancing are also considered to make sure a near optimal utilization of the computational resources as well as obtaining high quality output. A set of numerical experiments have demonstrated that our parallel implementation can not only reduce the execution time greatly but also obtain higher parallel efficiency.