Cache Based Multigrid on Unstructured Grids
نویسنده
چکیده
A Gauss-Seidel variant is developed which maintains data in the L2 cache memory longer than and runs approximately twice as fast as standard implementations. This variant depends on a decomposition of grid nodes into blocks which t into cache. We discuss two O(n) algorithms which perform a one-time reordering of the grid nodes and associated operators. Numerical tests demonstrate the speedups possible. A performance analysis tool connrms that our version makes signiicantly better use of L2 cache than standard versions.
منابع مشابه
Preprocessing Costs of Cache Based Multigrid
Multigrid algorithms based on cache aware smoothers produce a high performance, portable, cache aware multigrid solver for problems with one or more degrees of freedom on unstructured grids. This paper analyzes the work cost of a necessary preprocessing step in cache aware multigrid. This mesh renumbering step is not present in standard multigrid codes. The cost is found to be small relative to...
متن کاملCache Based Multigrid on Unstructured Two Dimensional Grids
1. Abstract. High speed cache memory is commonly used to address the disparity between the speed of a computer's central processing unit and the speed of a computer's main memory. It is advantageous to maximize the amount of time that data spends in cache. Tiling is a software technique which is often used to do just this. Tiling is not able, however, to handle dynamically changing data structu...
متن کاملCache Optimization for Structured and Unstructured Grid Multigrid∗
Many current computer designs employ caches and a hierarchical memory architecture. The speed of a code depends on how well the cache structure is exploited. The number of cache misses provides a better measure for comparing algorithms than the number of multiplies. In this paper, suitable blocking strategies for both structured and unstructured grids will be introduced. They improve the cache ...
متن کاملUsing Sparse Tiling with Symmetric Multigrid
Good data locality is an important aspect of obtaining scalable performance for multigrid methods. However, locality can be difficult to achieve, especially when working with unstructured grids and sparse matrices whose structure is not known until runtime. Our previous work developed full sparse tiling, a runtime reordering and rescheduling technique for improving locality. We applied full spa...
متن کاملAgglomeration Multigrid and Parallel Processing for the Numerical Solution of the Euler Equations on 2d and 3d Unstructured Grids
Abstract. This paper aims at coupling two known CFD techniques, namely multigrid and parallelization, in an existing Euler equations solver for 2D and 3D unstructured grids. The solver is based on a time-marching formulation for the high-subsonic/transonic flow equations and a pointwise implicit solution algorithm. The gain from the combined use of multigrid and parallelization is the reduction...
متن کامل