Towards Textbook Efficiency for Parallel Multigrid
نویسندگان
چکیده
In this work, we extend Achi Brandt’s notion of textbook multigrid efficiency (TME) to massively parallel algorithms. Using a finite element based geometric multigrid implementation, we recall the classical view on TME with experiments for scalar linear equations with constant and varying coefficients as well as linear systems with saddle-point structure. To extend the idea of TME to the parallel setting, we give a new characterization of a work unit (WU) in an architecture-aware fashion by taking into account performance modeling techniques. We illustrate our newly introduced parallel TME measure by large-scale computations, solving problems with up to 200 billion unknowns on a TOP-10 supercomputer. AMS subject classifications: 65N55, 68W10
منابع مشابه
A quantitative performance analysis for Stokes solvers at the extreme scale
This article presents a systematic quantitative performance analysis for large finite element computations on extreme scale computing systems. Three parallel iterative solvers for the Stokes system, discretized by low order tetrahedral elements, are compared with respect to their numerical efficiency and their scalability running on up to 786 432 parallel threads. A genuine multigrid method for...
متن کاملTextbook Multigrid Efficiency for Fluid Simulations∗
■ Abstract Recent advances in achieving textbook multigrid efficiency for fluid simulations are presented. Textbook multigrid efficiency is defined as attaining the solution to the governing system of equations in a computational work that is a small multiple of the operation counts associated with discretizing the system. Strategies are reviewed to attain this efficiency by exploiting the fact...
متن کاملTextbook Multigrid Efficiency for the Incompressible Navier-Stokes Equations: High Reynolds Number Wakes and Boundary Layers
Textbook multigrid efficiencies for high Reynolds number simulations based on the incompressible Navier-Stokes equations are attained for a model problem of flow past a finite flat plate. Elements of the Full Approximation Scheme multigrid algorithm, including distributed relaxation, defect correction, and boundary treatment, are presented for the three main physical aspects encountered: enteri...
متن کاملAIAA 99-3334 Distributed Relaxation Multigrid and Defect Correction Applied to the Compressible Navier-Stokes Equations
The distributed-relaxation multigrid and defectcorrection methods are applied to the twodimensional compressible Navier-Stokes equations. The formulation is intended for high Reynolds number applications and several applications are made at a laminar Reynolds number of 10,000. A staggeredgrid arrangement of variables is used; the coupled pressure and internal energy equations are solved togethe...
متن کامل