An automated post-allocation translator for parallel dynamic system simulation

The authors describe the post-allocation translator developed at the University of Alabama to transform allocation and modeling data into the source-level files necessary to execute parallel dynamic simulations within a multi-transputer-based computer architecture. Parallel multiple-instruction-multiple-data (MIMD) processing of large-scale dynamic system simulations can provide levels of performance which are not obtainable within purely sequential environments. A need exists for the creation of an environment which supports the easy amalgamation of parallel methods with the problem to be simulated. Ongoing research is described of the development of such an environment which automatically translates a simulation represented in a generic form into a parallel representation based on the selected parallel simulation methodology. An initial evaluation of the translator is discussed