Date of Award
1994
Document Type
Dissertation
Degree Name
Doctor of Philosophy (PhD)
Department
Computer Science
First Advisor
J. Bush Jones
Abstract
A supercompiler is a program that attempts to automatically restructure serial code into an equivalent parallel form. This restructuring is achieved through the application of various transformation strategies designed to remove data dependences. A data dependence is a relation between two programming statements that prevent those two statements from being executed in parallel. This research develops a rules based system to analyze the various data dependent transformation strategies of a supercompiler for parallel computers. With the information obtained from user input and the automated analysis of a program segment, this rules based analysis will be able to determine which of the available transformation strategies is the optimal one to be applied for a particular program segment.
Recommended Citation
Mcguffee, James Woodson, "A Rules Based Approach to Analyze Data Dependent Transformation Strategies of a Supercompiler for Parallel Computers." (1994). LSU Historical Dissertations and Theses. 5817.
https://repository.lsu.edu/gradschool_disstheses/5817
Pages
83
DOI
10.31390/gradschool_disstheses.5817