Scaling Commercial Verification to Larger Systems

Abstract

Simulation test coverage does not scale gracefully with growing system design size. Component interactions grow exponentially with the number of system components, while conventional system test at best can increase coverage as a linear function of allotted test time.

Likewise, capacity limitations are commonly cited as the essential gating factor that restricts the application of automatic formal verification (model checking) to at most a few design blocks.

Nonetheless, abstraction has long been used successfully in pilot projects to apply model checking to entire systems. Abstraction in conjunction with guided-random simulation can be used in the same way to increase coverage for conventional test.

While academic use of abstraction is old, its use in the EDA industry’s commercial tool sets has been very limited, due to a perception that its use entails an unacceptably disruptive methodology change. It is shown here how quite general data-path abstraction incorporated into a hierarchical design flow can be introduced with only a modest change in methodology. This hierarchical design flow supports verification based on either simulation or model checking that can scale gracefully with increasing design complexity.