报告 1：Oscillator-Based Computing, AMS Verification, Memristive Device Modelling, Random

             Telegraph Noise and Circuit/System Simulation: An Overview of Our Research

报告人：Professor Jaijeet Roychowdhury,

             EECS Department, University of California, Berkeley

            （Jaijeet Roychowdhury教授有意招募学生，欢迎有兴趣申请的同学到场交流）

摘   要：We provide glimpses of our ongoing research. In the PHLOGON project, we are developing novel implementations of computing primitives. By employing self-sustaining oscillators that encode logic values in phase, PHLOGON offers superior noise immunity and, potentially, energy efficiency by exploiting emerging nano-technologies. Our Accurate Booleanization of Continuous Systems (ABCD) project is a new approach to the long-standing analog/mixed-signal verification problem. By replacing continuous-time blocks (eg, SPICE-level circuits) by finite state machine approximations generated automatically by algorithm, we are able to bring scalable techniques for purely Boolean verification to bear on AMS systems. Our Berkeley Eye Estimator (BEE) identifies worst-case eyes in modern communication links without optimism or pessimism. We have developed the first suite of models for memristive devices (including RRAM) that are well posed mathematically and simulate properly in circuits. Our MUSTARD algorithm enables device-level Random Telegraph Noise to be co-simulated in a statistically correct manner with circuits. Finally, we have developed the Model and Algorithm Prototyping Platform (MAPP), which removes a long-standing barrier to research in models/algorithms for continuous-time systems: the lack of a powerful yet convenient platform for prototyping new device models and simulation algorithms quickly.

报告 2：Boolean Computation using Oscillators

报告人：Tianshi Wang,

             EECS Department, University of California, Berkeley

摘   要：Let's consider a fundamentally different paradigm for computing: instead of representing bits and bytes using voltage levels, we encode them in the phase of oscillatory signals. Phase encoding has been used for decades in radio communication for its superior noise immunity; we show that it can also be used in digital computation with virtually any type of self-sustaining nonlinear oscillators as the underlying logic elements. Such oscillators can come from not only CMOS technology, but also biology such as neurons, nanotechnology such as Spin Torque Nano-Oscillators (STNO) and MEMS resonators, and optics such as lasers. Several of them have great potential in high-speed and low-power operation, making oscillator-based computation a promising alternative to explore at the end of Moore's Law. In this talk, I will present the key concept, mechanism and working prototypes of oscillator-based Boolean computation. I will also discuss the challenges in the design and analysis of such computation systems, and sketch out our efforts in addressing them through the implementation of multiphysics devices and phase-macromodels in MAPP --- the Berkeley Model and Algorithm Prototyping Platform.