Survey of Different Data Dependence Analysis Techniques

( Vol-2,Issue-7,July 2016 )

Author(s): Monali Patil, Vandana Jagtap

Total View : 966
Downloads : 164
Page No: 1017-1022
DOI:

Keywords:

Dependence analysis, Parallelization, Data dependence test, Nonlinear subscript, Variable bound.

Abstract:

Dependency analysis is a technique to detect dependencies between tasks that prevent these tasks from running in parallel. It is an important aspect of parallel programming tools. Dependency analysis techniques are used to determine how much of the code is parallelizable. Literature shows that number of data dependence test has been proposed for parallelizing loops in case of arrays with linear subscripts, however less work has been done for arrays with nonlinear subscripts. GCD test, Banerjee method, Omega test, I-test dependence decision algorithms are used for one-dimensional arrays under constant or variable bounds. However, these approaches perform well only for nested loop with linear array subscripts. The Quadratic programming (QP) test, polynomial variable interval (PVI) test, Range test are typical techniques for nonlinear subscripts. The paper presents survey of these different data dependence analysis tests.

Cite This Article:

Share: