Parallel QR factorization using the torus-wrap mapping (Q1319516)
From MaRDI portal
 | This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use this page instead for the normal view: Parallel QR factorization using the torus-wrap mapping |
scientific article; zbMATH DE number 550158
| Language | Label | Description | Also known as |
|---|---|---|---|
| English | Parallel QR factorization using the torus-wrap mapping |
scientific article; zbMATH DE number 550158 |
Statements
Parallel QR factorization using the torus-wrap mapping (English)
0 references
12 April 1994
0 references
The QR factorization method is parallelized for multiprocessor systems with distributed memory. The Householder reflections are used for the matrix reduction in such a manner that the torus-wrap-mapping of data to processors can be adopted. The main contribution of the paper is a successful combination of both approaches which leads to an efficient execution of the communication pattern of the algorithm. Computational tests have been performed on the hypercube nCUBE-2 system with 1024 nodes.
0 references
computational tests
0 references
QR factorization
0 references
multiprocessor systems with distributed memory
0 references
Householder reflections
0 references
matrix reduction
0 references
torus- wrap-mapping
0 references
algorithm
0 references
0.8252317905426025
0 references
0.8180925250053406
0 references
0.8067382574081421
0 references
