Memory access problems in block cyclic reduction on vector computers

DOI10.1016/S0167-8191(05)80117-6MaRDI QIDQ806949zbMATH OpenFDO

Authors Hartmut Schwandt

Publication date 1991

Published in Parallel Computing (Search for Journal in Brave)

Full work available at URL https://doi.org/10.1016/S0167-8191(05)80117-6

zbMATH Keywords

block cyclic methodsblock tridiagonal coefficient matricesinterval arithmeticlinear systemsMemory accessvector computers

Mathematics Subject Classification ID

Direct numerical methods for linear systems and matrix inversion (65F05) Interval and finite arithmetic (65G30) Parallel numerical computation (65Y05)

Review text

The paper is concerned with linear systems with block tridiagonal coefficient matrices, which are also allowed to be of interval type. Generalizations of block cyclic methods as they were introduced by \textit{O. Bunnemann} [A compact noniterative Poisson solver, Inst. for Plasma Research, Report 294, Stanford University (1969)] are combined with interval arithmetic techniques as solution methods. Memory access problems are intended to be reduced drastically by the use of an appropriate data organization avoiding the usual powers of 2 as increments.

Recommendations

• Cyclic Reduction for Tridiagonal Systems of Equations with Interval Coefficients on Vector Computers
• Base-\(p\)-cyclic reduction for tridiagonal systems of equations
• Cyclic block-algorithms for solving triangular systems on distributed-memory multiprocessors with mesh topology
• A fast vector algorithm for solving tridiagonal linear equations
• Solving tridiagonal systems of linear equations on the IBM 3090 VF

This page was built for publication: Memory access problems in block cyclic reduction on vector computers

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q806949)