|
Informatics Report Series
|
|
|
|
|
|
|
|
|
| Title:Towards an Adaptive Java Optimising Compiler: An Empirical Evaluation of Program Transformations |
| Authors:
Shun Long
; Michael O'Boyle
|
| Date:Jun 2001 |
| Publication Title:Procs of 3rd Workshop on Java for High Performance Computing, ACM ICS |
|
Publication Type:Conference Paper
|
|
|
|
|
- Abstract:
-
Our overall aim is to develop an Adaptive Java Optimising Framework(AJOF) which dynamically optimises Java programs during execution. This paper focuses on one aspect of such a framework and investigates the impact of high level transformations on program performance. It shows that by combining transformations, it is possible to achieve speedups on a uni-processor of upto 1.86. This is followed by an evaluation of an adaptive feedback directed search strategy where it is is shown that it is possible to achieve 95% of the available performance with just 15 evaluations. Thus good transformation may be found relatively quickly and can be included in a on-line adaptive compilation system.
- Links To Paper
- 1st Link
- Bibtex format
- @InProceedings{EDI-INF-RR-0501,
- author = {
Shun Long
and Michael O'Boyle
},
- title = {Towards an Adaptive Java Optimising Compiler: An Empirical Evaluation of Program Transformations},
- book title = {Procs of 3rd Workshop on Java for High Performance Computing, ACM ICS},
- year = 2001,
- month = {Jun},
- url = {http://www.dcs.ed.ac.uk/home/mob/java01.ps},
- }
|
