Xin Li Carnegie Mellon University

High-Dimensional Strongly-Nonlinear VLSI Performance Modeling and Optimization for Nano-Scale Technologies

The aggressive scaling of CMOS technologies results in high-dimensional strongly-nonlinear performance variations that cannot be accurately captured by linear or even quadratic response surface models. These large-scale variations make it continually more challenging to create reliable and robust IC designs. In this talk, I will present several novel performance modeling techniques to facilitate robust IC design with affordable computational cost. Our preliminary results demonstrate that strongly-nonlinear response surface modeling for high-dimensional variation space (10K~100K parameters) is possible by applying advanced numerical algorithms.

Bio:

Xin Li received a Ph.D. degree in ECE from Carnegie Mellon University in 2005. He is currently a research scientist in the ECE Department at Carnegie Mellon University. He received the IEEE William J. McCalla ICCAD Best Paper Award in 2004, and in 2006 won both the Best Session Award from the SRC Student Symposium and the Best Paper Nomination from DAC. His research interests include modeling, simulation and synthesis for analog and digital systems.