Automatic Generation of High Quality Assertions

ECE Seminar: Automatic Generation of High Quality Assertions


Starts at: May 2, 2013 4:30 PM

Ends at: 5:30 PM

Location: Pittsburgh Campus: Scaife Auditorium (Room 125), Refreshments at 4pm. Silicon Valley campus: Rm 118, Building 23 at 1:30 PST

Speaker: Shobha Vasudevan

Affiliation: Dept of Electrical and Computer Engineering, University of Illinois at Urbana-Champaign

Link to Abstract

Details:

Abstract:

Assertions are ubiquitous in the design of today's systems due to their versatile roles in verification, reliability and analysis. The effectiveness of assertion based methodologies is highly contingent on the type, number and quality of assertions that are written. Knowing what assertions to write is a manual, time-intensive, ad-hoc, subjective process. In this talk, I will present methods for automatic assertion generation that rely on statistical as well as static analyses of the system. GoldMine, our assertion generation software, uses algorithms that combine data mining and control flow graph analysis to provide high quality assertions. I will present  real case studies in hardware and systems-on-a-chip where GoldMine has been successfully applied . I will also present algorithms that make GoldMine's assertions concise, precise and expressive.

I will also present a methodology for generating test inputs iteratively using GoldMine. The methodology generates a test set with monotonically increasing coverage, and achieves test coverage closure within a few iterations. 

I will finally discuss what a "good assertion" means- can we come up with a figure of merit for system properties? Is there an underlying principle of quality in the properties we express as humans? Can we objectify these notions for system properties?

Bio:
Shobha Vasudevan is an assistant professor at the Electrical and Computer Engineering (ECE) Department at the University of Illinois at Urbana-Champaign. She obtained her M.S. and PhD degrees in ECE from the University of Texas at Austin in 2004 and 2008 respectively. Her research interests are in hardware and embedded system verification and validation, software testing, hardware reliability and analog verification. Her research software GoldMine has been licensed by several semiconductor and EDA companies. She was a recipient of the NSF CAREER award in 2010 and the YWCA award for mentoring women in 2011. She has been named as the recipient of the ACM SIGDA Outstanding New Faculty award in 2013.