Understanding the parallel scalability of an implicit unstructured mesh CFD code
William Gropp, Dinesh Kaushik, David Elliot Keyes, Barry Smith · University of North Texas Digital Library (University of North Texas) · 2000
In this paper, the authors identify the scalability bottlenecks of an unstructured grid CFD code (PETSc-FUN3D) by studying the impact of several algorithmic and architectural parameters and by examining different programming models. The authors discuss the basic performance characteristics of this PDE code with the help of simple performance models developed in their earlier work, presenting primarily experimental results. In addition to achieving good per-processor performance (which has been addressed in the cited work and without which scalability claims are suspect) they strive to improve the implementation and convergence scalability of PETSc-FUN3D on thousands of processors.