Pregled bibliografske jedinice broj: 986260
Residual Replacement in Mixed-Precision Iterative Refinement for Sparse Linear Systems
Residual Replacement in Mixed-Precision Iterative Refinement for Sparse Linear Systems // ISC High Performance 2018: High Performance Computing / Yokota, Rio ; Weiland Michèle ; Shalf, John ; Alam, Sadaf (ur.).
Cham: Springer, 2018. str. 554-561 doi:10.1007/978-3-030-02465-9_39
CROSBI ID: 986260 Za ispravke kontaktirajte CROSBI podršku putem web obrasca
Naslov
Residual Replacement in Mixed-Precision Iterative Refinement for Sparse Linear Systems
Autori
Anzt, Hartwig ; Flegar, Goran ; Novaković, Vedran ; Quintana-Ortí, Enrique S. ; Tomás, Andrés E.
Vrsta, podvrsta i kategorija rada
Poglavlja u knjigama, znanstveni
Knjiga
ISC High Performance 2018: High Performance Computing
Urednik/ci
Yokota, Rio ; Weiland Michèle ; Shalf, John ; Alam, Sadaf
Izdavač
Springer
Grad
Cham
Godina
2018
Raspon stranica
554-561
ISBN
978-3-030-02464-2
ISSN
0302-9743
Ključne riječi
Sparse linear systems ; Krylov solvers Iterative refinement ; Mixed precision ; Residual replacement ; Performance and energy modelling
Sažetak
We investigate the solution of sparse linear systems via iterative methods based on Krylov subspaces. Concretely, we combine the use of extended precision in the outer iterative refinement with a reduced precision in the inner Conjugate Gradient solver. This method is additionally enhanced with different residual replacement strategies that aim to avoid the pitfalls due to the divergence between the actual residual and the recurrence formula for this parameter computed during the iteration. Our experiments using a significant part of the SuiteSparse Matrix Collection illustrate the potential benefits of this technique from the point of view, for example, of energy and performance.
Izvorni jezik
Engleski
Znanstvena područja
Matematika, Računarstvo
Citiraj ovu publikaciju:
Časopis indeksira:
- Scopus
