The Directed approach is easy to set up and is precise in terms of generating test suites to cover the target parameter combinations. However, it can quickly become labor intensive as it requires considerable manual work to generate many test suites for each of the features.

The Constrained Random approach automatically generates test suites. However, it is a wasteful methodology by nature, and as the coverage nears the target, its “hit rate” significantly worsens. This condition results in requiring substantially more time than truly needed.

Having to use both Directed and Constrained Random approaches is not efficient, to say the least. I, a nearly 20-year veteran in the register-transfer level design verification space, view it as a pain point.

This situation motivated me to start developing a tool with a new approach—an approach that has (i) the precision to hit the test target, (ii) the efficiency of automation, and (iii) the ability to provide traceability and visibility in the entire verification process. I set these three very clear goals, and then I started the development project in early 2016. The first production version was released in late 2018. Since then, a few field trials have been conducted on actual verification projects, and the tool has proven to improve productivity substantially.

In this paper, I discuss how my new test-generation tool, called iTGV (intelligent test generator for verification), can improve productivity in the CDV processes.................Download to read more