Sparse linear systems occur in areas such as finite element methods and statistics. These systems are often solved on parallel computers due to their size. In this paper a theoretical analysis of parallel sparse QR factorisation using a multifrontal method is undertaken. The analysis is quantified by some estimates of parallel speeds up for various parallel computers. These estimates show that only moderate parallel speedups can be attained.
@article{631,
title = {The scalability of parallel computers for sparse QR factorisation},
journal = {ANZIAM Journal},
volume = {42},
year = {2000},
doi = {10.21914/anziamj.v42i0.631},
language = {EN},
url = {http://dml.mathdoc.fr/item/631}
}
Miron, David J. The scalability of parallel computers for sparse QR factorisation. ANZIAM Journal, Tome 42 (2000) . doi : 10.21914/anziamj.v42i0.631. http://gdmltest.u-ga.fr/item/631/