An application level parallelization of complex real-time software

Abstract

Modern hardware infrastructure of complex computer based systems is based on multiprocessors and/or multi cores. Therefore, application software has to be executed in parallel on many processors/cores rather than sequentially on a single processor computer. This paper contributes to the overall research effort in the area by proposing an original approach to application level parallelization of complex real time software. In our previous work on parallelization of legacy software we have used the Task Tree Executor, a runtime engine for parallelization of Distribution Management System application functions. In this paper we present another solution, in the form of an architecture, which parallelizes a set of Distribution Management System application functions, controls their execution, and improves robustness of the system. The software architecture presented in this paper is named the Calculation Engine. The paper also presents the results of the performance evaluation based on a series of experiments made on the dual-core symmetric multiprocessor and NUMA machines. © 2010 IEEE.