Affordable Access

Towards verifying large(r) systems: A strategy and an experiment

Authors
Publisher
IFIP Lecture Notes in Computer Science (LNCS)
Publication Date
Disciplines
  • Computer Science
  • Design
  • Mathematics

Abstract

Despite some of the impressive results quoted in recent verification literature, the verification of even modestly sized "real" industrial designs is not yet feasible on a routine basis. The goal of the work discussed here is to enable the verification of large(r) real systems than currently feasible with any one of the available techniques, and to dovetail the verification methodology with the an underlying design methodology. The specific design methodology considered here is targeted towards the custom design of digital signal processing architectures. Two important attributes of this class of designs are (1) custom-crafted leaf cells, and (2) significant data path components. Our strategy is to partition the subsystems involved into different categories that can be handled by different techniques, and use a divide-and-conquer paradigm. The complexity introduced by data paths is addressed by automating the abstraction of some of the natural equivalences induced, and exploiting this in the context of an extended finite state machine formalism. Specifically, we illustrate how it is possible to exploit the distinction between data and control in implicitly specified state machines (HDLs), and comment on useful abstractions in the presence of symmetry. We illustrate some aspects of the strategy via an example. We also suggest some evolutionary (rather than revolutionary) changes in the design methodology that enable the existing state of art in verification to be better exploited in practise.Full Text at Springer, may require registration or fee

There are no comments yet on this publication. Be the first to share your thoughts.