[ 目次 ]
[ Q&A ]
SS研HPCフォーラム2006「次世代HPC技術が拓く大規模科学技術計算」
High Performance Computing and Parallel Numerical Algorithms in Computational Science and Engineering
アブストラクト
Supercomputing has proved to be an essential tool for advancing science and engineering disciplines. Applications in these disciplines range from computational nanoelectronics to bioinformatics and computational biology. Realizing high performance in these Computational Science and Engineering (CSE) applications, however, requires not only advanced compilers and software tools but also the design of those parallel algorithms and corresponding kernels that exploit best the underlying architecture and system software. In this presentation we offer examples of algorithm design of solvers for those algebraic linear systems and eigenvalue problems that arise in a variety of CSE applications. Also, we comment on current benchmarking procedures that "rank" a given parallel computing platform in terms of the maximum achievable peak performance, and assess its suitability for a given class of applications.
|
|