An approach of the QR factorization for tall-and-skinny matrices on multicore platforms

Authors:
Sergey V. Kuznetsov
Affiliations:
Software and Services Group, Intel Corporation, Novosibirsk, Russia
Venue:
PARA'12 Proceedings of the 11th international conference on Applied Parallel and Scientific Computing
Year:
2012

Citing 5
Cited 0

LAPACK's user's guide

LAPACK's user's guide
Orthogonal reduction of dense matrices to bidiagonal form on computers with distributed memory architectures

Parallel Computing
A Storage Efficient WY Representation for Products of Householder Transformations

A Storage Efficient WY Representation for Products of Householder Transformations
Comparative study of one-sided factorizations with multiple software packages on multi-core hardware

Proceedings of the Conference on High Performance Computing Networking, Storage and Analysis
Applying recursion to serial and parallel QR factorization leads to better performance

IBM Journal of Research and Development

Quantified Score

Hi-index	0.00

Visualization

Abstract

In this paper we focus primarily on a technique used to parallelize the LAPACK QR factorization of tall-and-skinny matrices. The modifications of the panel QR factorization we suggest neither affect the accuracy nor increase memory consumption. Results for tall-and-skinny matrices on the Intel® Xeon® platforms, and comparisons between the Intel® Math Kernel Library (Intel MKL) QR, PLASMA QR and the method proposed are provided.