Compiler Analysis of Sparse and Irregular Computations

We have studied five different possible parallelization methods for irregular reduction loops, all of which can be applied automatically by a compiler. We compared their ease of use, applicability, supporting compiler techniques required, run-time resource requirement, and, most importantly, run-time performance. From our analysis and experiments, we developed a general guideline for choosing an efficient parallel irregular reduction method for given programs and input data sets.

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

• Lin, Yuan

Contributors dc:contributor

• Padua, David A.

• Engineering, Electronics and Electrical

Language dc:language

eng

Identifier

(MiAaPQ)AAI9971122

OAI identifier oai:identifier

oai:www.ideals.illinois.edu:2142/81980