|M.Sc Student||Kiper Ron|
|Subject||Hole Analysis Coverage Using BDD'S|
|Department||Department of Electrical and Computers Engineering||Supervisor||PROF. Yoram Moses|
Functional Coverage is used to monitor the quality of the verification process, and its reports are used to help users identify areas in the system that have not yet been adequately tested. As complexity and size of the verified systems grows, stronger coverage tools are needed, that will be able to cope with bigger testing environments, and provide shorter more comprehensive reports. Hole Analysis is a modern method of presenting coverage, providing the user only with the most important practical information of the areas still uncovered within the testing environment, rather than a full coverage view.
A Reduced Ordered Binary Decision Diagram (ROBDD) is a powerful data structure for binary information due to its ability to successfully compact large-scale data, and efficiently operate some basic functions on them.
Since Hole Analysis does not require storing numeric information about the coverage events, and rather just their occurrence, this paper suggests using ROBDDs for solving the Hole Analysis problem.
Initially all collected coverage events are stored in an accumulating ROBDD, taking advantage of the strength of the database in compactly storing binary information. Retrieving the holes out of the collected events is later straightforward in ROBDDs, giving high performance to the analyzing of the Hole Analysis view. Despite some limitations that using ROBDDs for coverage raise, the overall performance of well defined coverage models will be shown to be very high on a variety of examples tested from hardware, as well as software, giving a better platform for good and fast coverage.