Source Transformation of Pure Parallel Regions
In this chapter, we will first describe the notation that we use for describing a parallel code region and its execution. This work is about the source transformation of pragma-based parallel regions. Therefore, we will define the SPL language that is used as input and as output language for our source transformation. After we have defined the source transformation rules, we will show that the resulting code represents the AD model as defined in the introduction chapter. The SPL language is shaped such that it comprises one compiler directive in form of a pragma that defines an associated code region as concurrently executable. We will denote this kind of parallel region as a pure parallel region.
Unable to display preview. Download preview PDF.