John Hayes

TITLE:  "Soft Errors Make Hard Problems for Chip Designers"

December 4, 2008, 3 p.m.

Abstract: The steady shrinking of the transistors used in integrated circuit (IC) chips makes them increasingly vulnerable to transient upsets caused by external radiation such as cosmic rays, or by internal effects such as electrical noise. Usually no permanent damage is done to the chips, but the resulting soft errors may lead to serious system failures. Moreover, many of the nanotechnologies being proposed to replace or supplement conventional ICs have behavior, both normal and faulty, that is inherently prone to random errors. To deal with such errors requires probabilistic methods that are not easily incorporated into the traditional chip design flow. In this talk, I will review the sources and impact of soft errors. I will describe some recent methods to model them and analyze their effects on digital circuit behavior and reliability. In particular, I will discuss a computational framework based on probabilistic transfer matrices (PTMs) to estimate soft error rates, noting the central role played by the tensor product and related algebraic operations. Finally, I will discuss some design techniques to make circuits fault-tolerant in the presence of soft errors.

Bio: John P. Hayes is Professor of EECS at the University of Michigan, Ann Arbor where he holds the Claude E. Shannon Chair of Engineering Science. His current teaching and research interests are in the areas of computer-aided design and testing, VLSI circuits, fault-tolerant architectures, and quantum computing. He received the B.E. degree from the National University of Ireland, and his M.S. and Ph.D. from the University of Illinois, all in electrical engineering. Prior to joining the University of Michigan, he was a faculty member at the University of Southern California. He has authored numerous technical papers, several patents, and five books, including a widely used text on computer architecture. He received the Michigan’s Distinguished Faculty Achievement Award in 1999 and the Humboldt Foundation’s Research Award in 2004. Professor Hayes is a Fellow of both IEEE and ACM.