Optimization through recomputation in the polyhedral model

Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

149 Downloads (Pure)

Abstract

Many modern (mobile) systems involve memory intensive computations. External memory accesses are costly when it comes to the execution time and energy consumption of a program. To overcome this, we usually apply tiling to improve data locality and data reuse in internal memories. In the research reported in this paper we add the possibility to recompute data rather than storing temporary results, and demonstrate that this can have a positive e ect on the overall application performance.
To achieve this we represented recomputation in the Polyhedral model by extending Polly. We experimentally veri ed the e ectiveness of recomputation on a pair of Convolutional Neural Network layers, when applying loop tiling, loop fusion, and recompute.
Original languageEnglish
Title of host publicationEighth International Workshop on Polyhedral Compilation Techniques
Subtitle of host publicationIn conjunction with HiPEAC 2018
Number of pages9
Publication statusPublished - 22 Jan 2018
Event8th International Workshop on Polyhedral Compilation Techniques (IMPACT 2018), January 23, 2018, Manchester, UK - Manchester, United Kingdom
Duration: 23 Jan 201823 Jan 2018
Conference number: 2018
http://impact.gforge.inria.fr/impact2018

Workshop

Workshop8th International Workshop on Polyhedral Compilation Techniques (IMPACT 2018), January 23, 2018, Manchester, UK
Abbreviated titleIMPACT
CountryUnited Kingdom
CityManchester
Period23/01/1823/01/18
OtherIn conjunction with HiPEAC 2018, January 22-24, 2018
Internet address

Fingerprint

Dive into the research topics of 'Optimization through recomputation in the polyhedral model'. Together they form a unique fingerprint.

Cite this