Data dependency analysis using data-write detection techniques

One of the most researched topics in parallel programming is data dependency analysis and data write detection analysis. Traditionally researchers have used mathematical models to identify conflicts in a complex code. The accuracy of such results, however, depends on the correctness of the mathematical model. Additionally, implementation of a mathematical model is a huge task by itself. In this paper, we discuss three techniques for data dependency analysis namely duplicate variables, page fault interrupt, and debug registers. The techniques are discussed in details using sample C programs. However, these techniques can be extended to other programming languages as well. We further discuss the merits and demerits of each of the techniques.