Enhancing the Stress and Efficiency of RIS Tools Using Coverage Metrics

Abstract

Random instruction sequence (RIS) tools continue to be the main strategy for verifying and validating chip designs. In every RIS tool, test suites are created targeted to a particular functionality and run on the design. Coverage metrics provide us one mechanism to ensure and measure the completeness and thoroughness of these test suites and create new test suites directed towards unexplored areas of the design. The results from the coverage metrics can also be used to improve the cluster efficiency. In this work we discuss the results from a coverage tool that extracted and analyzed stimuli quality from large regressions, using statistical visualization. Using this coverage tool, we captured events relating to the memory sub-system and improved the stress/efficiency of the tool by making the required modifications to the tool. We ran several experiments based on the event collection and increased the ability in the tool to create scenarios exercising patterns that can potentially highlight complex bugs.