Tutorial T1B: Emerging Computational Devices, Architectures and Computational Models

摘要

Summary form only given, as follows. The complete presentation was not made available for publication as part of the conference proceedings. This course will look at the synergies that are required across the stack from new devices to new computational models for designing future computing systems. This course will enable attendees to understand the inter-twined nature of design optimization that requires one to interact with experts in different domains. The students will be exposed to simulation tools and modelling techniques to help explore new circuits and architectures. As physical dimensional scaling alone has ceased to be the key factor driving the industry, many innovations have occurred in designing new types of logic switches including changes to their structure (such as three-dimensional FinFETs), their underlying physics (use of tunneling for steep-switching devices), the material systems (integration of ferro-electrics in gate stack for Negative Capacitance FETs). The first part of the lecture will introduce these devices, simulation models and accompanying circuit innovations. There has been a world of revolution in the memory devices with the emergence of many nonvolatile memory technologies and their tight integration in cross-point architectures. These memory systems have enabled new styles of computing systems such as the non-volatile processor for internet of thing systems and neuromorphic computing systems for cognitive computing. The lectures will focus on the synergistic coordination in advances in devices to system design. Neuromorphic systems also leverage new advances in technology such as cross-point memory arrays to integrate computing and store. Another emerging novel computational model is based on the principle “let physics do the computation”. This technique focuses on using the intrinsic operation mechanism of devices (such as nanoscale electronic coupled oscillators) to do the computation, instead of building complex circuits with standard transistors to carry out the same function. The primary objective is to train the next generation researchers and practitioners that can understand the synergy across the stack from devices to applications. This will prepare the next generation workforce for the beyond Moore era using post-CMOS devices and new computational paradigms beyond Von-Neumann computing models. This course will introduce the following topics: [1] Emerging logic and memory devices: What value do they add for circuit designers/architects? [2] Circuit/Architecture design using Emerging Logic and Memory devices [3] Neuromorphic and Brain-Inspired Computing using emerging devices [4] Computing Using Coupled Dynamical Systems.