Process Algebraic - Performance Modeling of Embedded Software

More Info
expand_more

Abstract

A compiler for embedded platforms has many optimization flags providing code size and speed improvement. Traditional profiling methods take lot of time to identify the best combination of the compiler flags to suit the requirement especially if the software stack is very huge. AUTOSAR is one such growing software market in which there is a need for rapid performance assessment. In this thesis a means to estimate the performance of a program using the process algebraic language (PEPA) is investigated. The assembly program is converted in to the PEPA model and the performance measures obtained by solving the model is verified against the actual execution time of the program. The experimental results provide valuable insights on the methodology.