Selection of Fixed-Point or Floating-Point Autocoding for Production Applications

Lecturers: Lev Vitkin, Susan Dong, Chandrashekar MS, Delphi Electronics & Safety

Abstract:
Since the end of the 1990s, the model-based development (MBD) process has increasingly been adopted for the development of automotive embedded control software. Although this process reduces the time for algorithm development via creation of the executable models, significant time is still required for the software implementation of these models. The MBD approach opened the way for the generation of code automatically from the executable model and the deployment of that code directly to the production controllers. This paper focuses on comparison of the development times for production-intent, fixed-point autocode against floating-point autocode for the models that are developed in MATLAB^®/Simulink^®/Stateflow^®. The reductions in the development time for the floating-point software could offset the cost increase of the hardware due to the need for higher-end, floating-point microprocessors for the execution of the generated software. The paper’s studies are based on Delphi experience with fixed-point and floating-point autocoding for the production applications.