A Quantitative Approach to Functional Debugging

Functional debugging dominates both development time and cost of modern design process. Two dominant debugging techniques are simulation and emulation. Neither of them provides an adequate debugging solution. Design simulation is two to ten orders of magnitude slower than functional execution with respect to the fabricated design. Design emulation and fabrication provide high execution speed, but also demonstrate very serious disadvantage: they signiicantly restrict the observability and controllability of majority of variables in the design. To facilitate debugging, in particular error detection, we introduce a novel cut-based functional debugging paradigm. It coordinates design emulation and simulation and enables fast transition from one to another. Therefore, we leverage on the advantages of both debugging domains in order to provide fast functional debugging as well as full observability and controllability of all variables. Emulation or functional implementation is used for fast application execution during debugging, while simulation provides complete observability and controllability. The eeective implementation of the new debugging approach poses several CAD tasks. We formulate the optimization tasks, establish their complexity, and develop least-constraining most-constrained heuristics to solve them. The eeciency and eeectiveness of the new approach and accompanying algorithms is demonstrated on a set of designs where full observability and con-trollability are accomplished with low hardware overhead.