The combining DAG: a technique for parallel data flow analysis

The availability of multiprocessor systems has recently generated an increased interest in the design and implementation of parallel compilers. The authors present a parallel algorithm for performing data flow analysis which is an important component of any compiler. Data flow information is used in various phases of compilation including code generation, code scheduling, traditional optimizing transformations and parallelizing transformations. A single program compilation may involve the computation of the solution to several data flow problems and possibly solutions to the same problem multiple times at different points in the language translation process. Thus, for large programs the data flow analysis performed during compilation can be time-consuming. Data flow analysis is also used in other program tools such as editors, testers, and debuggers.<<ETX>>