1Out-of-Core Shared Memory Parallel Delaunay Mesh Renement

Abstract-We present an out-of-core algorithm for Parallel Delaunay Renement (PDR) for shared memory machines. Our per-formance data on a small 4-way shared memory machine in-dicate that the out-of-core PDR algorithm is 6.5 times faster than the in-core PDR implementation that relies on operating system virtual memory for meshes larger than 30 million el-ements. However, as it was expected, it is about twice slower than the in-core PDR implementation for meshes smaller than 20 million elements