Xin Li Carnegie Mellon University

Projection-Based Performance Modeling for Statistical Circuit Analysis and Optimization

Large-scale process fluctuations in nano-scale IC technologies suggest applying high-order (e.g., quadratic) response surface models to capture the circuit performance variations. Applying such nonlinear models requires significantly more characterization time and results in non-Normal performance distribution. In this talk, we propose several novel techniques (e.g. PROBE, APEX) to address this nonlinearity problem and make an optimal tradeoff between modeling accuracy and computational cost. The accuracy and efficiency of the proposed techniques are demonstrated by various large-size problems (10K+ random process parameters including mismatches).

Bio:

Xin Li received the Ph.D. degree in ECE from Carnegie Mellon University in 2005. He was a summer intern at Extreme DA, Palo Alto, CA, in 2004. In the summer of 2005, he joined the faculty of the Dept. of ECE, Carnegie Mellon University. His current research interests include modeling, simulation and synthesis for analog/RF and digital systems. Dr. Xin Li received the IEEE/ACM William J. McCalla ICCAD Best Paper Award in 2004.