Published January 1988
| public
Book Section - Chapter
Optimal matrix algorithms on homogeneous hypercubes
- Other:
- Fox, Geoffrey
Abstract
This paper describes a set of concurrent algorithms for matrix algebra, based on a library of collective communication routines for the hypercube. We show how a systematic application of scattering reduces load imbalance. A number of examples are considered (Gaussian elimination, Gauss-Jordan matrix inversion, the power method for eigenvectors, and tridiagonalisation by Householder's method), and the concurrent efficiencies are discussed.
Additional Information
© 1988 ACM. The support of the Department of Energy, under grant number DE-FG03-85ER25009, is gratefully acknowledged. The initial study of the Householder algorithm was due to T. Delbruck, and we would like to thank Paul Hipes for emphasising the relevance of the Gauss-Jordan algorithm.Additional details
- Eprint ID
- 71422
- DOI
- 10.1145/63047.63125
- Resolver ID
- CaltechAUTHORS:20161024-165545288
- DE-FG03-85ER25009
- Department of Energy (DOE)
- Created
-
2016-10-25Created from EPrint's datestamp field
- Updated
-
2021-11-11Created from EPrint's last_modified field