TUES, THU 11 AM- 12.30 PM CST
VIRTUAL (Online) CLASS MEETINGS
INSTRUCTOR: PROF. SHOBHA VASUDEVAN
shobhav AT [this university]
Office hours: Friday 4:00 PM-6:00 PM
COURSE OVERVIEW
Details of the course outline can be found here.
With every modern feature we desire in our systems like low power, high speed, high functionality, great form factor etc. the tradeoff is in their verification complexity. Verification is the biggest bottleneck in hardware and embedded system development utilizing about 70% of time and resources in the system design cycle. In other words, we build systems we do not know how to verify.
This course introduces the most scalable and efficient verification algorithms researched in the past 30 years and used widely in contemporary industry. This course teaches algorithms for pre-Silicon verification with full observability into the system, and post-Silicon validation with limited observability after chip manufacture. Due to the rising prominence of analog and mixed signal circuits in modern systems, the course briefly covers advances in analog and mixed signal verification. This course exposes the student to all aspects of the verification phase of system design, including algorithms for automatic invariant generation, debug, and diagnosis. Coverage metrics for evaluating the extent of system verification are critical to practical verification, and are discussed. The course also addresses verification of hardware systems with underlying uncertainty in behavior through statistical guarantees. The algorithms taught include formal verification algorithms like equivalence checking, model checking, symbolic simulation etc. as well as simulation based verification algorithms. Students will have a chance to design algorithms for, and use state-of-the-art verification tools in the course.
This course is intended for (1) graduate students looking to pursue research in verification/test/post-Silicon (2) graduate students who want to apply verification to their research (3) graduate students to learn scalable algorithms that can solve complex search problems. In general, this provides an insight into a critically important aspect of system design that contemporary industry highly values.