Year: 2006
Type: article
Abstract: The implementation of statically and dynamically optimised virtual machines for resource constrained devices does not appear easy without trade offs between functionality, performance and memory use.Java virtual machines that interpret Java bytecode appear to be the answer, but are inherently slow.Avenues have been explored that can help with this, all of which appear to be equally viable.Our attention is turned towards the actual driving force behind the machine (the instruction set) and shows that through the use of simple, well established methodology's, a more compact instruction set can be derived for the Java virtual machine through the analysis of static properties of application data.This is desirable for a number of reasons allowing us to optimise implementations of virtual machines without explicitly modifying the way in which a Java virtual machine works.This dissertation describes the implementation of a system for extracting extensive statistics from a number of Java applications, and makes use of the statistics to describe a number of possible amendments to the existing Java virtual machine instruction set.
Author: Alan Shaw
Language: English
Subfield: Software
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W120699731