Informatics Report Series


Report   

EDI-INF-RR-0458


Related Pages

Report (by Number) Index
Report (by Date) Index
Author Index
Institute Index

Home
Title:Automatic Tuning of Inlining Heuristics
Authors: John Cavazos ; Michael O'Boyle
Date:Nov 2005
Publication Title:International Conference for High Performance Computing, Networking, and Storage (SC|05), Seattle WA
Publisher:IEEE Computer Society
Publication Type:Conference Paper Publication Status:Published
Page Nos:13
DOI:10.1109/SC.2005.14 ISBN/ISSN:1-59593-06
Abstract:
Inlining improves the performance of programs by reducing the overhead of method invocation and increasing the opportunities for compiler optimization. Incorrect inlining decisions, however, can degrade both the running and compilation time of a program. This is especially important for a dynamically compiled language such as Java. Therefore, the heuristics that control inlining must be carefully tuned to achieve a good balance between these two costs to reduce overall total execution time. This paper develops a genetic algorithms based approach to automatically tune a dynamic compilers internal inlining heuristic. We evaluate our technique within the Jikes RVM [1] compiler and show a 17% average reduction in total execution time on the SPECjvm98 benchmark suite on a Pentium-4. When applied to the DaCapo benchmark suite, our approach reduces total execution time by 37%, outperforming all existing techniques.
Links To Paper
1st Link
Bibtex format
@InProceedings{EDI-INF-RR-0458,
author = { John Cavazos and Michael O'Boyle },
title = {Automatic Tuning of Inlining Heuristics},
book title = {International Conference for High Performance Computing, Networking, and Storage (SC|05), Seattle WA},
publisher = {IEEE Computer Society},
year = 2005,
month = {Nov},
pages = {13},
doi = {10.1109/SC.2005.14},
url = {http://www.anc.ed.ac.uk/machine-learning/colo/inlining.pdf},
}


Home : Publications : Report 

Please mail <reports@inf.ed.ac.uk> with any changes or corrections.
Unless explicitly stated otherwise, all material is copyright The University of Edinburgh