Myers Research Group

University of Utah



People

Research

Tools

Publications

Book

Home Page


System-Level Timing Analysis and Verification

SUPPORTED BY

SRC-2002-TJ-1024: System-Level Timing Verification With Automatic Abstraction

SRC-99-TJ-694: Timing Verification using Automatic Abstraction

SRC-97-TJ-487.001: Verification of Timed Circuits

NSF Japan Program Award INT-0087281

NSF CAREER award MIP-9625014



The goal of this research task is to develop methodology and tools to support system-level timing verification. Systems are composed of a heterogeneous collection of digital, mixed signal/analog, and software components. As a result, their state space tends to be very large. Therefore, to be practical this tool must support automatic abstraction. Our group has developed one of the most efficient timing verification tools for circuit-level analysis. This tool has been utilized by companies such as Intel and IBM for analysis of their aggressively optimized high performance circuits. Our verification tool was used during the design of several of the critical circuits in the Intel RAPPID design. Our tool has also been used to verify several circuits from the IBM guTS processor which is pictured above. Most recently, we developed automatic abstraction techniques that allowed our tool to be used to efficiently verify several variations of IBM's Synchronous Interlocked Pipelines.

The objective of this project is to extend our tools applicability to the system-level. Key to this will be developing methods for specifying properties of a system composed of heterogeneous components in a high level functional model. This model must be related formally to lower level structural descriptions so that properties of the whole system can be verified efficiently. To achieve efficient verification of such systems, techniques to automatically abstract the environment to support assume/guarantee methods must be developed. Our primary interest is in verifying timing properties of these systems. The addition of real-time complicates reachability analysis, so we will need to continue to produce even more efficient timed state space reachability analysis methods. We plan to further improve our abstraction and verification algorithms by targeting the properties that are to be verified. Finally, it is likely that exhaustive methods cannot be applied to many large systems, so we will also investigate bounded model checking techniques with our goal of finding the most bugs possible. When exhaustive methods cannot be applied, we plan to also provide coverage metrics to give the designer feedback on the quality of the verification result. In our previous research, our collaborations with industry have been essential to our success. These companies provided motivating examples for our timing verification tool that helped not only validate the utility of our tool, but also to provide future research directions. In this project, we plan to continue to use examples from industry to drive the direction of this research.

Faculty:

Students:

Alumni:

Books:

PhD Theses:

Master's Theses:

Bachelor's Theses:

Journal Papers:

Conference Papers:

Workshop Papers: