Degree Type

Thesis

Date of Award

2012

Degree Name

Master of Science

Department

Electrical and Computer Engineering

First Advisor

Glenn R Luecke

Abstract

Unied Parallel C (UPC) is a language used to write parallel programs for shared and distributed memory parallel computers. UPC-CHECK is a scalable tool developed to automatically detect argument errors in UPC functions and deadlocks in UPC programs at run-time and issue high quality error messages to help programmers quickly x those errors. The tool is easy to use and involves merely replacing the compiler command with upc-check. The tool uses a novel distributed algorithm for detecting argument and deadlock errors in collective operations. The run-time complexity of the algorithm has been proven to be O(1). The algorithm has been extended to detect deadlocks created involving locks with a run-time complexity of O(T), where T is the number of threads waiting to acquire a lock. Error messages issued by UPC-CHECK were evaluated using the UPC RTED test suite for argument errors

in UPC functions and deadlocks. Results of these tests show that the error messages issued by UPC-CHECK for these tests are excellent. The scalability of all the algorithms used was demonstrated using performance-evaluation test programs and the UPC NAS Parallel Benchmarks.

DOI

https://doi.org/10.31274/etd-180810-1846

Copyright Owner

Indranil Roy

Language

en

Date Available

2012-10-31

File Format

application/pdf

File Size

77 pages

Share

COinS