QR factorization of a dense matrix on a hypercube multiprocessor
In this article we describe a new algorithm for computing the QR factorization of a rectangular matrix on a hypercube multiprocessor. The scheme involves the embedding of a two-dimensional grid in the hypercube network. We employ a global communication scheme which uses redundant computation to maintain data proximity, and the mapping strategy is such that for a fixed number of processors the processor idle time is small and either constant or grows linearly with the dimension of the matrix. A complexity analysis tells us what the aspect ratio of the embedded grid should be in terms of the shape of the matrix and the relative speeds of communication and computation. Numerical experiments performed on an Intel Hypercube multiprocessor support the theoretical results. 21 refs., 20 tabs.
- Research Organization:
- Waterloo Univ., Ontario (Canada). Dept. of Computer Science; Oak Ridge National Lab., TN (USA)
- DOE Contract Number:
- AC05-84OR21400
- OSTI ID:
- 5254478
- Report Number(s):
- ORNL/TM-10691; ON: DE88009187
- Country of Publication:
- United States
- Language:
- English
Similar Records
QR factorization of a dense matrix on a shared-memory multiprocessor
An efficient communication scheme for solving the S{sub n} equations on message-passing multiprocessors
Related Subjects
HYPERCUBE COMPUTERS
ALGORITHMS
MATRICES
FACTORIZATION
COMMUNICATIONS
EFFICIENCY
PERFORMANCE TESTING
COMPUTERS
MATHEMATICAL LOGIC
TESTING
990230* - Mathematics & Mathematical Models- (1987-1989)
990210 - Supercomputers- (1987-1989)