Circuits and Interconnects In Aggressively Scaled CMOS

Slides:



Advertisements
Similar presentations
Topics Electrical properties of static combinational gates:
Advertisements

Elettronica T A.A Digital Integrated Circuits © Prentice Hall 2003 Inverter CMOS INVERTER.
Introduction to CMOS VLSI Design Lecture 21: Scaling and Economics
11/29/2004EE 42 fall 2004 lecture 371 Lecture #37: Memory Last lecture: –Transmission line equations –Reflections and termination –High frequency measurements.
CS294-6 Reconfigurable Computing Day4 September 3, 1998 VLSI Scaling.
Institute of Digital and Computer Systems 1 Fabio Garzia / Finding Peak Performance in a Process23/06/2015 Chapter 5 Finding Peak Performance in a Process.
S. Reda EN160 SP’07 Design and Implementation of VLSI Systems (EN0160) Lecture 22: Material Review Prof. Sherief Reda Division of Engineering, Brown University.
1 Modeling and Optimization of VLSI Interconnect Lecture 1: Introduction Avinoam Kolodny Konstantin Moiseev.
Introduction to CMOS VLSI Design Lecture 21: Scaling and Economics Credits: David Harris Harvey Mudd College (Material taken/adapted from Harris’ lecture.
Computation Energy Randy Huang Sep 29, Outline n Why do we care about energy/power n Components of power consumption n Measurements of power consumption.
The CMOS Inverter Slides adapted from:
6.893: Advanced VLSI Computer Architecture, September 28, 2000, Lecture 4, Slide 1. © Krste Asanovic Krste Asanovic
THE INVERTERS. DIGITAL GATES Fundamental Parameters l Functionality l Reliability, Robustness l Area l Performance »Speed (delay) »Power Consumption »Energy.
1 CS/EE 6810: Computer Architecture Class format:  Most lectures on YouTube *BEFORE* class  Use class time for discussions, clarifications, problem-solving,
Modern VLSI Design 2e: Chapter 3 Copyright  1998 Prentice Hall PTR Topics n Electrical properties of static combinational gates: –transfer characteristics;
Low Power – High Speed MCML Circuits (II)
The End of Conventional Microprocessors Edwin Olson 9/21/2000.
Basics of Energy & Power Dissipation
© Digital Integrated Circuits 2nd Inverter Digital Integrated Circuits A Design Perspective The Inverter Jan M. Rabaey Anantha Chandrakasan Borivoje Nikolic.
Trends in IC technology and design J. Christiansen CERN - EP/MIC
Analog to Digital Converters
FPGA-Based System Design: Chapter 6 Copyright  2004 Prentice Hall PTR Topics n Low power design. n Pipelining.
Scaling I Mohammad Sharifkhani. Reading Text book II (pp. 123)
Caltech CS184 Winter DeHon 1 CS184a: Computer Architecture (Structure and Organization) Day 6: January 22, 2003 VLSI Scaling.
Penn ESE370 Fall DeHon 1 ESE370: Circuit-Level Modeling, Design, and Optimization for Digital Systems Day 6: September 19, 2011 Restoration.
EE 587 SoC Design & Test Partha Pande School of EECS Washington State University
Seok-jae, Lee VLSI Signal Processing Lab. Korea University
Penn ESE534 Spring DeHon 1 ESE534 Computer Organization Day 19: March 28, 2012 Minimizing Energy.
CS203 – Advanced Computer Architecture
Penn ESE370 Fall DeHon 1 ESE370: Circuit-Level Modeling, Design, and Optimization for Digital Systems Day 30: November 21, 2012 Crosstalk.
Penn ESE370 Fall DeHon 1 ESE370: Circuit-Level Modeling, Design, and Optimization for Digital Systems Day 6: September 17, 2012 Restoration.
Power-Optimal Pipelining in Deep Submicron Technology
CS203 – Advanced Computer Architecture
Temperature and Power Management
Lecture 05: Static Behaviour of CMOS Inverter
VLSI Testing Lecture 5: Logic Simulation
Low-Power SRAM Using 0.6 um Technology
Stateless Combinational Logic and State Circuits
VLSI Testing Lecture 5: Logic Simulation
Hot Chips, Slow Wires, Leaky Transistors
Basics of Energy & Power Dissipation
Very low voltage 16-bit counter in high leakage static CMOS technology
Day 29: November 18, 2011 Dynamic Logic
Registers and clocking issues
Architecture & Organization 1
VLSI Design MOSFET Scaling and CMOS Latch Up
Downsizing Semiconductor Device (MOSFET)
Reading: Hambley Ch. 7; Rabaey et al. Sec. 5.2
Day 6: September 11, 2013 Restoration
Day 33: November 19, 2014 Crosstalk
Day 22: October 31, 2011 Pass Transistor Logic
Architecture & Organization 1
Digital Integrated Circuits 17: CMOS III: Design and Scaling
Day 17: October 18, 2010 (Energy) Ratioed Logic
Day 31: November 23, 2011 Crosstalk
Day 23: November 2, 2012 Pass Transistor Logic: part 2
CS/EE 6810: Computer Architecture
Downsizing Semiconductor Device (MOSFET)
Dual Mode Logic An approach for high speed and energy efficient design
Circuit Design Techniques for Low Power DSPs
Circuit Characterization and Performance Estimation
Estimating Delays Would be nice to have a “back of the envelope” method for sizing gates for speed Logical Effort Book by Sutherland, Sproull, Harris Chapter.
A High Performance SoC: PkunityTM
Day 25: November 7, 2011 Registers
Day 21: October 29, 2010 Registers Dynamic Logic
Day 5: September 17, 2010 Restoration
Technology scaling Currently, technology scaling has a threefold objective: Reduce the gate delay by 30% (43% increase in frequency) Double the transistor.
Lecture 15: Scaling & Economics
Lecture 15: Scaling & Economics
A Quasi-Delay-Insensitive Method to Overcome Transistor Variation
Presentation transcript:

Circuits and Interconnects In Aggressively Scaled CMOS 9/6/2018 Circuits and Interconnects In Aggressively Scaled CMOS Mark Horowitz Computer Systems Laboratory Stanford University horowitz@stanford.edu VLSI Scaling

Device Scaling In digital CMOS design: Only two circuit forms matter (maybe three) Static CMOS, and Dynamic CMOS These forms are used because: They don’t demand much from devices So they work with crummy transistors Robust, especially static circuits VLSI Scaling

FO4 Inverter Delay Under Scaling Device performance will scale FO4 delay has been linear with tech Approximately 0.36 nS/mm*Ldrawn at TT (0.5nS/mm under worst-case conditions) Easy to predict gate performance We can measure them Labs have built 0.04mm devices Key issue is voltage scaling 1 3 5 7 0.5 1.5 2 FO4 delay (nS) Feature size (mm) 0.36 * Ldrawn VLSI Scaling

Circuit Power Is very much tied to voltage scaling If the power supply scales with technology For a fixed complexity circuit Power scales down as a^3 if you run as same frequency Power scales down as a^2 if you run it 1/ a times faster Power scaling is a problem because Freq has been scaling at faster than 1/ a Complexity of machine has been growing This will continue to be an issue in future chips Remember scaling the technology makes a chip lower power! VLSI Scaling

Voltage Scaling Circuits performance depends on the Vdd to Vth ratio Ideally both should scale together If Vth scales leakage scales If Vth does not scale, gates get slower, or Vdd can’t scale as fast and power goes up Leakage is easier to deal with than power, transistors will leak Tech 0.8 0.5 0.35 0.25 0.18 0.13 0.10 Vdd 5 3.3 2.5 1.8 1.3 1.0 Vth ?? VLSI Scaling

Scaling Global Wires R gets quite a bit worse with scaling; C basically constant 0.1 0.2 0.3 0.4 0.25 0.18 0.13 0.07 0.05 0.035 Kohms Technology Ldrawn (um) Semi-global wire resistance, 1mm long Aggressive scaling Conservative scaling 0.2 0.4 0.6 0.25 0.18 0.13 0.1 0.07 0.05 0.035 pF Technology Ldrawn (um) Semi-global wire capacitance, 1mm long Aggressive scaling Conservative scaling VLSI Scaling

Scaling Module Wires R is basically constant, and C falls linearly with scaling 0.1 0.2 0.3 0.4 0.25 0.18 0.13 0.07 0.05 0.035 Kohms Technology Ldrawn (um) Semi-global wire resistance, scaled length Aggressive scaling Conservative scaling 0.2 0.4 0.6 0.25 0.18 0.13 0.1 0.07 0.05 0.035 pF Technology Ldrawn (um) Semi-global wire capacitance, scaled length Aggressive scaling Conservative scaling VLSI Scaling

Architecture Scaling Plot of IPC Compiler + IPC 1.5x / generation What next? Wider machines Threads Speculation Guess answers to create parallelism Have high wire costs Won’t be easy VLSI Scaling

Clock Frequency Most of performance comes from clock scaling Clock frequency double each generation Two factors contribute: technology (1.4x/gen), circuit design VLSI Scaling

Gates Per Clock Clock speed has been scaling faster than base technology Number of FO4 delays in a cycle has been falling Number of gates decrease 1.4x each generation Caused by: Faster circuit families (dynamic logic) Better optimization Approaching a limit: <16 FO4 is hard < 8 FO4 is very hard VLSI Scaling