GPU Implementation of Finite Difference SolversReport as inadecuate




GPU Implementation of Finite Difference Solvers - Download this document for free, or read online. Document in PDF available to download.

Reference: Giles, M, Laszlo, E, Reguly, I et al., (2014). GPU Implementation of Finite Difference Solvers.Citable link to this page:

 

GPU Implementation of Finite Difference Solvers

Abstract: This paper discusses the implementation of one-factor and three-factor PDE models on GPUs. Both explicit and implicit time-marching methods are considered, with the latter requiring the solution of multiple tridiagonal systems of equations.Because of the small amount of data involved, one-factor models are primarily compute-limited, with a very good fraction of the peak compute capability being achieved. The key to the performance lies in the heavy use of registers and shuffle instructions for the explicit method, and a non-standard hybrid Thomas/PCR algorithm for solving the tridiagonal systems for the implicit solverThe three-factor problems involve much more data, and hence their execution is more evenly balanced between computation and data communication to/from the main graphics memory. However, it is again possible to achieve a good fraction of the theoretical peak performance on both measures. The high performance requires particularly careful attention to coalescence in the data transfers, using local shared memory for small array transpositions, and padding to avoid shared memory bank conicts.Computational results include comparisons to computations on Sandy Bridge and Haswell Intel Xeon processors, using both multithreading and AVX vectorisation.

Peer Review status:Peer reviewedPublication status:PublishedVersion:Accepted Manuscript Funder: Engineering and Physical Sciences Research Council   Conference Details: Proceedings of WHPCF 2014: 7th Workshop on High Performance Computational Finance - Held in conjunction with SC 2014: The International Conference for High Performance Computing, Networking, Storage and AnalysisNotes:Copyright © 2014 IEEE. Personal use of this material is permitted. Permission from IEEE must beobtained for all other uses, in any current or future media, includingreprinting/republishing this material for advertising or promotional purposes, creating newcollective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.

Bibliographic Details

Publisher: IEEE

Publisher Website: http://www.ieee.org/

Host: Proceedings of WHPCF 2014: 7th Workshop on High Performance Computational Finance - Held in conjunction with SC 2014: The International Conference for High Performance Computing, Networking, Storage and Analysissee more from them

Publication Website: http://ieeexplore.ieee.org/xpl/mostRecentIssue.jsp?punumber=7015707

Issue Date: 2014

pages:1-8Identifiers

Urn: uuid:3dd5fefc-1fd6-4d4b-a666-290ed668b991

Isbn: 9781479970278

Source identifier: 541656

Doi: https://doi.org/10.1109/WHPCF.2014.10 Item Description

Type: Conference;

Version: Accepted ManuscriptKeywords: computational finance GPU computing vectorisation tridiagonal equations Tiny URL: pubs:541656

Relationships





Author: Giles, M - institutionUniversity of Oxford Oxford, MPLS, Mathematical Inst - - - Laszlo, E - institutionUniversity of Oxford Oxfo

Source: https://ora.ox.ac.uk/objects/uuid:3dd5fefc-1fd6-4d4b-a666-290ed668b991



DOWNLOAD PDF




Related documents