Optimizing LLVM Pass Sequences with Shackleton: A Linear Genetic Programming Framework

Hannah Peeler, Shuyue Stella Li, Andrew N. Sloss, Kenneth N. Reid, Yuan Yuan, Wolfgang Banzhaf

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

5 Scopus citations

Abstract

In this paper we explore the novel application of a linear genetic programming framework, Shackleton, to optimizing sequences of LLVM optimization passes. The algorithm underpinning Shackleton is discussed, with an emphasis on the effects of different features unique to the framework when applied to LLVM pass sequences. Combined with analysis of different hyperparameter settings, we report the results on automatically optimizing pass sequences with Shackleton for two software applications at differing complexity levels. Finally, we reflect on the advantages and limitations of our current implementation and lay out a path for further improvements. These improvements aim to surpass hand-crafted solutions with an automatic discovery method for an optimal pass sequence.

Original languageEnglish
Title of host publicationGECCO 2022 Companion - Proceedings of the 2022 Genetic and Evolutionary Computation Conference
PublisherAssociation for Computing Machinery, Inc
Pages578-581
Number of pages4
ISBN (Electronic)9781450392686
DOIs
StatePublished - 9 Jul 2022
Externally publishedYes
Event2022 Genetic and Evolutionary Computation Conference, GECCO 2022 - Virtual, Online, United States
Duration: 9 Jul 202213 Jul 2022

Publication series

NameGECCO 2022 Companion - Proceedings of the 2022 Genetic and Evolutionary Computation Conference

Conference

Conference2022 Genetic and Evolutionary Computation Conference, GECCO 2022
Country/TerritoryUnited States
CityVirtual, Online
Period9/07/2213/07/22

Keywords

  • compiler optimization
  • evolutionary algorithms
  • genetic programming
  • metaheuristics
  • parameter tuning

Fingerprint

Dive into the research topics of 'Optimizing LLVM Pass Sequences with Shackleton: A Linear Genetic Programming Framework'. Together they form a unique fingerprint.

Cite this