The usefulness of Bounded Model Checking (BMC) based on propositional satisfiability (SAT) methods for bug hunting has already been proven in several recent work. In this paper, we present two industrial strength systems performing BMC for both verification and falsification. The first is Thunder, which performs BMC on top of a new satisfiability solver, SIMO. The second is Forecast, which performs BMC on top of a BDD package. SIMO is based on the Davis Logemann Loveland procedure (DLL) and features the most recent search methods. It enjoys static and dynamic branching heuristics, advanced back-jumping and learning techniques. SIMO also includes new heuristics that are specially tuned for the BMC problem domain. With Thunder we have achieved impressive capacity and productivity for BMC. Real designs, taken from Intel's Pentium © 4, with over 1000 model variables were validated using the default tool settings and without manual tuning. In Forecast, we present several alternatives for adapting BDD-based model checking for BMC. We have conducted comparison of Thunder and Forecast on a large set of real and complex designs and on almost all of them Thunder has demonstrated clear win over Forecast in two important aspects: capacity and productivity.

No abstract

In this paper, we describe a completely automated framework for iterative abstraction refinement that is fully integrated into a formal-verification environment. This environment consists of three basic software tools: Forecast, a BDD-based model checker, Thunder, a SAT-based bounded model checker, and MCE, a technology for multiple-counterexample analysis. In our framework, the initial abstraction is chosen relative to the property under verification. The abstraction is model checked by Forecast; in case of failure, a counterexample is returned. Our framework includes an abstract counterexample analyzer module that applies techniques for bounded model checking to check whether the abstract counterexample holds in the concrete model. If it does, it is extended to a concrete counterexample. This important capability is provided as a separate tool that also addresses one of the major problems of verification by manual abstraction. If the counterexample is spurious, we use a novel refinement heuristic based on MCE to guide the refinement. After the part of the abstract model to be refined is chosen, our refinement algorithm computes a new abstraction that includes as much logic as possible without adding too many new variables, therefore striking a balance between refining the abstraction and keeping its size manageable. We demonstrate the effectiveness of our framework on challenging Intel designs that were not amenable to BDD-based model-checking approaches.

No abstract

Our experience with semi-exhaustive verification shows a severe degradation in usability for the corner-case bugs, where the tuning effort becomes much higher and recovery from dead-ends is more and more difficult. Moreover, when there are no bugs at all, shifting semi-exhaustive traversal to exhaustive traversal is very expensive, if not impossible. This makes the output of semi-exhaustive verification on non-buggy designs very ambiguous. Furthermore, since after the design fixes each falsification task needs to converge to full verification, there is a strong need for an algorithm that can handle efficiently both verification and falsification. We address these shortcomings with an enhanced reachability algorithm that is more robust in detecting corner-case bugs and that can potentially converge to exhaustive reachability. Our approach is similar to that of Cabodi et al. in partitioning the frontiers during the traversal, but differs in two respects. First, our partitioning algorithm trades quality for time resulting in a significantly faster traversal. Second, the subfrontiers are processed according to some priority function resulting in a mixed BFS/DFS traversal. It is this last feature that makes our algorithm suitable for both falsification and verification.