JIT-Based Cost Analysis for Dynamic Program Transformations

Morton, J. M. , Maier, P. and Trinder, P. (2016) JIT-Based Cost Analysis for Dynamic Program Transformations. In: RAC2016, Eindhoven, Netherlands, 02 Apr 2016, 5 - 25. (doi: 10.1016/j.entcs.2016.12.012)

137422.pdf - Published Version
Available under License Creative Commons Attribution Non-commercial No Derivatives.



Tracing JIT compilation generates units of compilation that are easy to analyse and are known to execute frequently. The AJITPar project investigates whether the information in JIT traces can be used to dynamically transform programs for a specific parallel architecture. Hence a lightweight cost model is required for JIT traces. This paper presents the design and implementation of a system for extracting JIT trace information from the Pycket JIT compiler. We define three increasingly parametric cost models for Pycket traces. We determine the best weights for the cost model parameters using linear regression. We evaluate the effectiveness of the cost models for predicting the relative costs of transformed programs.

Item Type:Conference Proceedings
Additional Information:RAC 2016 - Resource Aware Computing
Keywords:Parallelism, resource analysis, cost model.
Glasgow Author(s) Enlighten ID:Morton, Mr John and Maier, Dr Patrick and Trinder, Professor Phil
Authors: Morton, J. M., Maier, P., and Trinder, P.
Subjects:Q Science > QA Mathematics > QA75 Electronic computers. Computer science
College/School:College of Science and Engineering > School of Computing Science
Research Group:GLASS
Journal Name:Electronic Notes in Theoretical Computer Science
Copyright Holders:Copyright © 2016 The Authors
First Published:First published in Electronic Notes in Theoretical Computer Science 330: 5-25
Publisher Policy:Reproduced under a Creative Commons License

University Staff: Request a correction | Enlighten Editors: Update this record

Project CodeAward NoProject NamePrincipal InvestigatorFunder's NameFunder RefLead Dept
644791Adaptive Just-In-Time Parallelisation (AJITPar)Phil TrinderEngineering & Physical Sciences Research Council (EPSRC)EP/L000687/1COM - COMPUTING SCIENCE