Kahan’s paranoia is a famous method for exercising the arithmetic of a normal computer. Unfortunately, its structure is not suited to the host / target architecture of embedded control systems. This paper describes a re-written and enhanced version of paranoia (ESP – Embedded System Paranaoia), which addresses this problem allowing them to be tested to the same degree of rigour as conventional systems. Given that many conventional systems fail paranoia to some extent, embedded systems are expected to behave similarly badly.
esp_19.zip (33kb)
Synopsis and invited feedback
This work was or is being reviewed by domain-specific experts appointed independently.
Invited Feedback
- Version 1.9: this includes excellent feedback from Adrian Capel and Simon Dalley of UKAEA Culham. Amongst other things, they have added the capability of dealing with machines which use extra internal precision more gracefully. I have added simple transcendental and hyperbolic function tests.
Related papers and links
esp.pdf (160kb)
