CIS 4930: Special Topics: Physical Limits of Computing.

Department of Computer and Information Science and Engineering

Computer Engineering technical elective (research survey course)

Catalog description

Survey of research on the physical limits and scaling properties of present and potential future computing technologies.  Fundamental bounds on computation from thermodynamics, relativity, and quantum mechanics; semiconductor technology scaling; energy-efficient "adiabatic" circuits; proposed future computing technologies such as quantum dots, superconducting logic, DNA computing, nanomechanical computing, molecular electronics, and quantum coherent computing; physics-based models of computation and asymptotic advantages of thermodynamically reversible circuits, architectures, programming languages, and algorithms.

Class/laboratory schedule

Credits: 3. Three 50 minute lecture/discussion sessions per week.

Prerequisites

In its initial, experimental phase, the course is open to students having a broad range of types and levels of experience.  However, students should at least have had some exposure to basic principles of computer science, digital logic, and/or computer engineering.

Textbooks and/or other required material

Required:
  • Numerous research articles, lecture notes, and slides, on web.
  • Frank, Reversibility for Efficient Computing, manuscript.

    • Recommended:
  • Hey (ed.), Allen (ed.), and Feynman.  Feynman Lectures on Computation.  Perseus, 1996.
  • Hey (ed.), Feynman and Computation: Exploring the Limits of Computers, Perseus, 1998.
  • Williams and Clearwater, Explorations in Quantum Computing, Springer Verlag, 1997.

    • Course objectives

    To familiarize students with state-of-the-art research on the potentialities and limitations of present and future computing technologies, based on principles of fundamental and applied physics.  This includes both near-term concerns with coming generations of semiconductor technology, and longer-term issues that will affect any possible future computing technology.  For example, students learn why the 2nd law of thermodynamics yields a requirement for thermodynamically and logically reversible processing in the most scalable possible computer architectures as device sizes approach the nanoscale, and further, how this type of processing can be implemented in today's VLSI technology for greater energy efficiency in portable and embedded systems.   Students will also learn about the tantalizing possibility of exponentially more powerful computing (on some problems) and provably secure communication that is emerging from research in quantum coherent computing.  They will also become familiar with a broad range of computing technologies that have been proposed to replace the MOSFET transistor when that device reaches its particular limits.

    Topics covered:

    - Fundamental physical constraints on computation including the speed of light limit, quantum limits on information density and processing rates, and the second law of thermodynamics.
    - The future of semiconductor technology, as forecasted by the International Technology Roadmap for Semiconductors and first-order semiconductor technology scaling laws.  Limits to scaling.
    - Energy-efficient "adiabatic" reversible logic: Basic principles; pipelined, sequential, fully-reversible logic circuits; power supply design issues.
    - Potential future computing technologies: Quantum dots & other nanoelectronic structures, superconductor electronics, quantum computing (& cryptographic implications), DNA computing, nanomechanical logic, molecular electronics
    - Physics-based models of computation: Desiderata, proposed models, scaling advantages of primarily-reversible models.
    - Reversible computing: General irreversible-to-reversible conversions, reversible serial and parallel architectures, reversible programming languages and algorithms.

    Contribution of this course to professional component

    In this interdisciplinary engineering course, the student learns how a wide range of  engineering disciplines (e.g. thermal engineering, electronic design, computer architecture, and algorithm design) interact in computer engineering, and furthermore that these disparate areas become increasingly inter-related as device sizes approach their fundamental physical limits.  The course treats design optimization of emerging computing systems not only under physical constraints but also (1) economic - We discuss cost factors including materials, area rental, and energy/cooling costs; and (2) manufacturing-related - we discuss how manufacturing costs scale with device size, and review the prospects for manufacturability of all the emerging technologies we discuss.

    Relationship to computer engineering program objectives

    Objective 1.  The course includes a review of core material on MOSFET transistor operating principles, CMOS logic gate designs, boolean logic circuits, and examples of computer architectures, programming languages and algorithms.
    Objective 2.  The course discusses and ties together nearly all elements of computer system design (except peripherals), including circuits, interconnections, architecture, compilers, physical packing and packaging, system geometry, power, and cooling.
    Objective 3.  The course discusses instruction-set design and implementation and hardware-software interactions, e.g., that for optimal efficiency, thermodynamically reversible hardware requires a new, logically reversible programming language, not just a new compiler for traditional languages.
    Objective 4.  The course makes extensive use of analytical techniques including asymptotic order-of-growth notation and algebraic solution of tradeoff/optimization problems.  It introduces students to design techniques for  novel types of digital circuits, computer architectures, and algorithms.

    Person(s) and Date

    Prepared by Dr. Michael P. Frank,  May 2000.