Introduction

Abstract

Modern integrated circuits and systems consist of many different functional blocks comprising multiple heterogeneous processor cores, dedicated analog/mixed signal components, various on-chip busses and memories, (third-party) Intellectual Property (IP), and most notably more and more embedded software. Following “Moore’s Law”, the available chip capacity grows exponentially. Currently, high-end processor designs reaches up to 2 billion transistors. A complete system can be integrated onto a single chip which is then called System-on-a-Chip (SoC). The increasing design complexity and scale of SoC designs combined with non-functional requirements and constraints on the final product, e.g. low power, robustness, reliability, and low cost, make the verification of the design correctness a complex and crucial task. Functional errors are still the most important cause of design respins. According to a study from Collett International Research [Cir04] nearly 40% of all chip designs require at least one re-spin. There, 75% of these designs contain functional or logical bugs. The increasing amount of embedded software implemented in integrated circuits further complicates verification. Studies, e.g. [Hum04], implicate that software still contains about 10 to 20 defects per 1,000 lines of code after compiling and testing is done. Remarkably, software companies have to spend nearly the same cost and time efforts on quality assurance like hardware manufacturers have to invest [Tas02].