Optimization of object-oriented and concurrent programs

High level programming language features have long been seen as improving programmer efficiency at some cost in program efficiency. When features such as object-orientation and fine-grained concurrency, which greatly simplify expression of complex programs, are used parsimoniously, their effectiveness is mitigated. It is my thesis that these features can be implemented efficiently through interprocedural analysis and transformation. By specializing their implementation to the contexts in which they are used, the program's efficiency is not adversely affected by the flexibility of the language. The specific contributions herein are: (1) an adaptive flow analysis for practical precise analysis of object-oriented programs, (2) a cloning algorithm for building specialized versions of general abstractions, (3) a set of optimizations for removing object-oriented and fine-grained concurrency overhead, and (4) a hybrid sequential-parallel execution model which adapts to the availability of data. The effectiveness of this framework has been empirically validated on standard benchmarks. It is publicly available as part of the Illinois Concert system (http://www-csag.cs.uiuc.edu).

Name thesis:degree_name

Ph.D.

Level thesis:degree_level

Dissertation

Discipline thesis:degree_discipline

Computer Science

Grantor

University of Illinois at Urbana-Champaign

Year dc:date

2011

Author dc:creator

• Plevyak, John Bradley

Contributors dc:contributor

• Chien, Andrew

• Computer Science

Statement dc:rights

• Copyright 1996 Plevyak, John Bradley

Language dc:language

eng

Identifier

9780591200379
AAI9712408
(UMI)AAI9712408

OAI identifier oai:identifier

oai:www.ideals.illinois.edu:2142/21582