Recursive least-squares using a hybrid Householder algorithm on massively parallel SIMD systems

Within the context of recursive least-squares, the implementation of a Hous eholder algorithm for block updating the QR decomposition, on massively par allel SIMD systems, is considered. Initially, two implementations based on different mapping strategies for distributing the data matrices over the pr ocessing elements of the parallel computer are investigated. Timing models show that neither of these implementations is superior in all cases. In ord er to increase computational speed, a hybrid implementation uses performanc e models to partition the problem into two subproblems which are then solve d using the first and second implementation, respectively. (C) 1999 Elsevie r Science B.V. All rights reserved.