A Motivating Case Study on Code Variant Selection by Reinforcement Learning

Titelangaben

Hacker, Oliver ; Korch, Matthias ; Seiferth, Johannes:
A Motivating Case Study on Code Variant Selection by Reinforcement Learning.
In: Varbanescu, Ana-Lucia ; Bhatele, Abhinav ; Luszczek, Piotr ; Marc, Baboulin (Hrsg.): High Performance Computing : Proceedings. - Cham : Springer , 2022 . - S. 293-312 . - (Lecture Notes in Computer Science ; 13289 )
ISBN 978-3-031-07312-0
DOI: https://doi.org/10.1007/978-3-031-07312-0_15

Angaben zu Projekten

Abstract

In this paper, we investigate the applicability of reinforcement learning as a possible approach to select code variants. Our approach is based on the observation that code variants are usually convertible between one another by code transformations. Actor-critic proximal policy optimization is identified as a suitable reinforcement learning algorithm. To study its applicability, a software framework is implemented and used to perform experiments on three different hardware platforms using a class of explicit solution methods for systems of ordinary differential equations as an example application.