Exploiting task and data parallelism in ILUPACK’s preconditioned CG solver on NUMA architectures and many-core accelerators
详细信息 查看全文
- 作者：José ; I. Aliaga ; a ; aliaga@icc.uji.es" class="auth_mail" title="E-mail the corresponding author ; Rosa M. Badiab ; rosa.m.badia@bsc.es" class="auth_mail" title="E-mail the corresponding author ; Maria Barredaa ; mvaya@icc.uji.es" class="auth_mail" title="E-mail the corresponding author ; Matthias Bollhö ; ferc ; m.bollhoefer@tu-bs.de" class="auth_mail" title="E-mail the corresponding author ; Ernesto Dufrechoud ; edufrechou@fing.edu.uy" class="auth_mail" title="E-mail the corresponding author ; Pablo Ezzattid ; pezzatti@fing.edu.uy" class="auth_mail" title="E-mail the corresponding author ; Enrique S. Quintana-Ortí ; a ; quintana@icc.uji.es" class="auth_mail" title="E-mail the corresponding author
- 关键词：Sparse linear systems ; Reconditioned Conjugate Gradient solver ; Task and data parallelism ; Multi-core processors ; Intel Xeon Phi ; Graphics processing units (GPUs)
- 刊名：Parallel Computing
- 出版年：2016
- 出版时间：May 2016
- 年：2016
- 卷：54
- 期：Complete
- 页码：97-107
- 全文大小：1505 K
文摘
- •
Specialized implementations of ILUPACK’s iterative solver for NUMA platforms.
- •
Specialized implementations of ILUPACK’s iterative solver for many-core accelerators.
- •
Exploitation of task parallelism via OmpSs runtime (dynamic schedule).
- •
Exploitation of task parallelism via MPI (static schedule).
- •
Exploitation of data parallelism for GPUs.