Year: 2000
Type: article
Abstract: . 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 usage without changing the underlying algorithm. In particular, bitwise compatibility is guaranteed between the standard and the high performance implementations of the algorithms. This is illustrated by comparisons for various multigrid algorithms on a selection of dierent computers for problems in two and three dimensions. The code restructuring can yield performance improvements of factors of 2-5. This allows the modied codes to achieve a much higher percentage of the peak performance of the CPU than is usually observed with standard implementations. Key words. Computer architectures, i...
Language: English
Subfield: Hardware and Architecture
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
Awards: A Fully Algebraic Multigrid Software Package for Automated Solution of Unstructured Sparse Linear Systems on High Performance Computers, Collaborative Research: High Performance Multi-Scale Ocean Modelling, Linear and Nonlinear Multigrid Methods on Cache Based Parallel and Serial Computers with Applications
OpenAlex ID: W1774469134
