Callahan and Kennedy [34] proposed distributed-memory compilation techniques based on data-dependence driven program transformations. These techniques were implemented in a prototype compiler in the ParaScope programming environment.
Under the ADAPT system [35] developed at the University of Southampton, the parallelization process of Fortran 90 is guided by distribution declarations for arrays, in a similar but, more restricted approach than Fortran D.
The ADAPTOR [36] is a tool that transforms data parallel programs written in Fortran with array extensions and layout directives to explicit message passing.
Li and Chen [38][37] describe general compiler optimization techniques that reduce communication overhead for Fortran-90 implementation on massively parallel machines. Our compiler uses similiar pattern matching techniques to Li and Chen to detect communication.