Faster coverage closure: Runtime guidance of Constrained Random stimuli by collected

Abstract

Constrained Random Verification (CRV) provides a great value towards faster coverage closure over Directed Tests Verification. Yet, closing coverage is not an easy procedure given current designs' sizes and complexities. Long simulation runs, huge amount of random tests, directed tests covering specific corner cases, and others, are all techniques being used to close coverage. One of the reasons for challenges in closing coverage is redundant stimuli, stemming from the lack of a feedback from collected coverage to Constrained Random (CR) stimuli generators at run time (e.g. during simulation), which could result in missing interesting stimuli. This paper shows a standard and efficient way to avoid redundant stimuli by building a link between collected coverage and CR stimuli at run time seeking faster coverage closure, i.e. how collected coverage can dynamically guide random stimuli generators during simulation to generate uncovered scenarios, and avoid generating redundant ones.