Autotuning of Adaptive Mesh Refinement PDE Solvers on Shared Memory Architectures

Handle URI:
http://hdl.handle.net/10754/597640
Title:
Autotuning of Adaptive Mesh Refinement PDE Solvers on Shared Memory Architectures
Authors:
Nogina, Svetlana; Unterweger, Kristof; Weinzierl, Tobias
Abstract:
Many multithreaded, grid-based, dynamically adaptive solvers for partial differential equations permanently have to traverse subgrids (patches) of different and changing sizes. The parallel efficiency of this traversal depends on the interplay of the patch size, the architecture used, the operations triggered throughout the traversal, and the grain size, i.e. the size of the subtasks the patch is broken into. We propose an oracle mechanism delivering grain sizes on-the-fly. It takes historical runtime measurements for different patch and grain sizes as well as the traverse's operations into account, and it yields reasonable speedups. Neither magic configuration settings nor an expensive pre-tuning phase are necessary. It is an autotuning approach. © 2012 Springer-Verlag.
Citation:
Nogina S, Unterweger K, Weinzierl T (2012) Autotuning of Adaptive Mesh Refinement PDE Solvers on Shared Memory Architectures. Lecture Notes in Computer Science: 671–680. Available: http://dx.doi.org/10.1007/978-3-642-31464-3_68.
Publisher:
Springer Science + Business Media
Journal:
Parallel Processing and Applied Mathematics
KAUST Grant Number:
UK-c0020
Issue Date:
2012
DOI:
10.1007/978-3-642-31464-3_68
Type:
Book Chapter
ISSN:
0302-9743; 1611-3349
Sponsors:
This publication is partially based on work supportedby Award No. UK-c0020, made by the King Abdullah University of Science andTechnology (KAUST). Computing resources for the present work have also beenprovided by the Gauss Centre for Supercomputing under grant pr63no.
Appears in Collections:
Publications Acknowledging KAUST Support

Full metadata record

DC FieldValue Language
dc.contributor.authorNogina, Svetlanaen
dc.contributor.authorUnterweger, Kristofen
dc.contributor.authorWeinzierl, Tobiasen
dc.date.accessioned2016-02-25T12:43:33Zen
dc.date.available2016-02-25T12:43:33Zen
dc.date.issued2012en
dc.identifier.citationNogina S, Unterweger K, Weinzierl T (2012) Autotuning of Adaptive Mesh Refinement PDE Solvers on Shared Memory Architectures. Lecture Notes in Computer Science: 671–680. Available: http://dx.doi.org/10.1007/978-3-642-31464-3_68.en
dc.identifier.issn0302-9743en
dc.identifier.issn1611-3349en
dc.identifier.doi10.1007/978-3-642-31464-3_68en
dc.identifier.urihttp://hdl.handle.net/10754/597640en
dc.description.abstractMany multithreaded, grid-based, dynamically adaptive solvers for partial differential equations permanently have to traverse subgrids (patches) of different and changing sizes. The parallel efficiency of this traversal depends on the interplay of the patch size, the architecture used, the operations triggered throughout the traversal, and the grain size, i.e. the size of the subtasks the patch is broken into. We propose an oracle mechanism delivering grain sizes on-the-fly. It takes historical runtime measurements for different patch and grain sizes as well as the traverse's operations into account, and it yields reasonable speedups. Neither magic configuration settings nor an expensive pre-tuning phase are necessary. It is an autotuning approach. © 2012 Springer-Verlag.en
dc.description.sponsorshipThis publication is partially based on work supportedby Award No. UK-c0020, made by the King Abdullah University of Science andTechnology (KAUST). Computing resources for the present work have also beenprovided by the Gauss Centre for Supercomputing under grant pr63no.en
dc.publisherSpringer Science + Business Mediaen
dc.titleAutotuning of Adaptive Mesh Refinement PDE Solvers on Shared Memory Architecturesen
dc.typeBook Chapteren
dc.identifier.journalParallel Processing and Applied Mathematicsen
dc.contributor.institutionTechnische Universitat Munchen, Munich, Germanyen
kaust.grant.numberUK-c0020en
All Items in KAUST are protected by copyright, with all rights reserved, unless otherwise indicated.