Compiling for Distributed Memory Multiprocessors Based on Access Region Analysis

This dissertation explores the applicability of fully automatic parallelizing techniques for distributed memory multiprocessors. In the research, an ordinary Fortran 77 program is assumed as input, and no information is required from the programmers. Only the abilities of the compiler are used to detect parallelism from the input program, to distribute data, and to control communication in the system. To achieve this goal, a variety of compiler techniques are used, including traditional compiling techniques and new techniques developed specifically with distributed memory architectures as the target. Combining these traditional and new techniques, various sequential benchmark programs are tested on one of the distributed memory machines.

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

2015

Author dc:creator

• Paek, Yunheung

Contributors dc:contributor

• Padua, David A.

• Computer Science

Language dc:language

eng

Identifier

(MiAaPQ)AAI9737215

OAI identifier oai:identifier

oai:www.ideals.illinois.edu:2142/81885