Leading Edge Logic Comparison. March 9, 2018 Scotten W. Jones President IC Knowledge LLC

Similar documents
Philips SAA7118E Multi Standard Video Decoder Process Review

VLSI Design 12. Design Styles

Same 10 Mask Makers in 2017, 2018 Surveys

Layout Design II. Lecture Fall 2003

Stack Height Analysis for FinFET Logic and Circuit

Designing of Low Power and Efficient 4-Bit Ripple Carry Adder Using GDI Multiplexer

CPE/EE 427, CPE 527 VLSI Design I L06: Complementary CMOS Logic Gates

ECE520 VLSI Design. Lecture 9: Design Rules. Payman Zarkesh-Ha

WHY CHINA IS A GOOD MARKET FOR FD SOI

Design of Low Power and High Speed 4-Bit Ripple Carry Adder Using GDI Multiplexer

Moving Towards Profitability

VLSI Design 14. Memories

Scaling. Krish Chakrabarty 1. Scaling

Comparing Indexes Among Primates

Synchronous Sequential Logic. Topics. Sequential Circuits. Chapter 5 Steve Oldridge Dr. Sidney Fels. Sequential Circuits

ENHANCED PARKWAY STUDY: PHASE 2 CONTINUOUS FLOW INTERSECTIONS. Final Report

save percentages? (Name) (University)

An Architecture for Combined Test Data Compression and Abort-on-Fail Test

VLSI Design I; A. Milenkovic 1

VLSI Design I; A. Milenkovic 1

MOSIS LAMBDA DESIGN RULES

Over the Edge. Forecasting beyond the Boundaries of the Regional Model

PX800 M E T A L PX800 PERFORMANCE WIL T-08

VLSI Design I; A. Milenkovic 1

COACH WORKBOOK. Contexts: Module: Outfield Play. Competition Development. Coaching Association of Canada, 2008 Baseball Canada 20011

PART 3 MODULE 6 GEOMETRY: UNITS OF GEOMETRIC MEASURE

EE241 - Spring 2013 Advanced Digital Integrated Circuits. Assigned reading. No new reading. Lecture 4: Transistor Models

For IEC use only. Technical Committee TC3: Information structures, documentation and graphical symbols

EE 434 Lecture 6. Process Technology

Summary report on all harvested species on Patuxent Research Refuge from September 1 - January 31, 2017 Deer Harvest

Probability of an Ace in Disc Golf by stevewestdiscgolf.com April 18, 2011

CPE/EE 427, CPE 527 VLSI Design I L21: Sequential Circuits. Review: The Regenerative Property

Technique Sheet 16. Using the Metric Ruler

VLSI Design I; A. Milenkovic 1

Guidelines for Providing Access to Public Transportation Stations APPENDIX C TRANSIT STATION ACCESS PLANNING TOOL INSTRUCTIONS

Introduction to Measurement Developing Standard and Metric Measuring Skills

Scanning Laser Vibrometry Assessment of Sports Equipment

HW #5: Digital Logic and Flip Flops

Mastering the Mechanical E6B in 20 minutes!

Modern and Ancient Tides

PX1 PX1 PERFORMANCE M E T A L WIL T-12

LECTURE 5 PRECEDENCE DIAGRAMING METHOD: LINK RELATIONSHIP

Stock Assessment Update for Pink Shrimp (Farfantepenaeus duorarum) in the U.S. Gulf of Mexico for 2015

Calibration Procedure

Stock Assessment Update for Pink Shrimp (Farfantepenaeus duorarum) in the U.S. Gulf of Mexico for 2014

STRUCTURED INQUIRY: Investigating Surface Area to Volume Ratio in Cells

Automating Injection Molding Simulation using Autonomous Optimization

Lab #1: Introduction to Lab Techniques INTRODUCTION

Experiment P18: Buoyant Force (Force Sensor)

where ρ f is the density of the fluid, V is the submerged volume of the object, and g is the acceleration due to gravity.

Common-duct series-cell electrolyzer with anomalous gas production rate

EUV Resist Performance under High Stray Light Levels: an Interference Lithography Study

Factor/Lowest Common Multiple

Use of Equivalent Building Dimensions (EBD) to Characterize Upwind Terrain Wake Effects for AERMOD

Performance Measure Summary - San Jose CA. Performance Measures and Definition of Terms

Performance Measure Summary - Chicago IL-IN. Performance Measures and Definition of Terms

DESIGN AND CHARACTERIZATION OF 20NM SOI MOSFET DOPING ABRUPTNESS DEPENDENT

Performance Measure Summary - Denver-Aurora CO. Performance Measures and Definition of Terms

Pipetting and Determining Protein Concentration

SPIRIT III Radiometer Saturation Effect

PX4 P L A S T I C PX4 PERFORMANCE WIL T-02

Economic Value of Celebrity Endorsements:

Preliminary design of a high-altitude kite. A flexible membrane kite section at various wind speeds

Exercise 3. Power Versus Wind Speed EXERCISE OBJECTIVE DISCUSSION OUTLINE. Air density DISCUSSION

Manufacturers Continue Capacity Expansion as Technology Orders Grow

Tokyo: Simulating Hyperpath-Based Vehicle Navigations and its Impact on Travel Time Reliability

GAME 1: 2 v 2 Why play this Game: Scenario What will this Game do? What do I need to set up this Game? What will pitch layout look like?

Freescale Semiconductor MMA6222AEG ±20/20g Dual-Axis Medium-g Micromachined HARMEMS Accelerometer

TRAFFIC STUDY GUIDELINES Clarksville Street Department

Global OSAT Market with Focus on IC Packaging: Size, Trends & Forecasts ( ) August 2018

Lab 13: Hydrostatic Force Dam It

Background Information. Project Instructions. Problem Statement. EXAM REVIEW PROJECT Microsoft Excel Review Baseball Hall of Fame Problem

Conflating a Traffic Model Network With a Road Inventory. David Knudsen

MnROAD Mainline IRI Data and Lane Ride Quality MnROAD Lessons Learned December 2006

Connecting Sacramento: A Trip-Making and Accessibility Study

Off-gassing from III-V Wafer Processing -- A Collaboration of SEMATECH and IMEC

Chapter 2 Displaying and Describing Categorical Data

Cover Page for Lab Report Group Portion. Drag on Spheres

Milking Systems, Selection, Cost and Implications

Multilane Roundabouts

6. EXPERIMENTAL METHOD. A primary result of the current research effort is the design of an experimental

Complete Wristband System Tutorial PITCHING

Copy of my report. Why am I giving this talk. Overview. State highway network

REFERENCE PAGE FIREARMS AND BALLISTICS PART 1: BULLETS AND BARRELS NAME

Tips for trainers. Participatory quota setting

#LZ400 LEAKALYZER. Water Loss Sensor INSTRUCTION MANUAL Country Dr. #190 St. Paul, MN

PX400 P L A S T I C PX400 PERFORMANCE WIL T-02

Physical Design of CMOS Integrated Circuits

Chapter 12: Mechanical Waves and Sound

Engineering Practice on Ice Propeller Strength Assessment Based on IACS Polar Ice Rule URI3

Performance Metrics Using Laser Operator and Laser Approval Surveys

Planning & Zoning Commission

Chapter 13 Fluids. Copyright 2009 Pearson Education, Inc.

INTERNATIONAL HYDROGRAPHIC SURVEY STANDARDS

Designing Wave Energy Converting Device. Jaimie Minseo Lee. The Academy of Science and Technology The Woodlands College Park High School, Texas

Technical Bulletin 148. Comparison of Flat Panel to MiniPanel. by:

Alternative Methods For Making CoCoRaHS Snow Water Content Measurements. Prepared by: Rick Fleetwood Environment Canada Updated Feb 2015

Adiabatic Switching. A Survey of Reversible Computation Circuits. Benjamin Bobich, 2004

Deirdre Boelke, NEFMC Staff, Scallop PDT Chair. Inshore Scallop Workshop February 22-23, 2016

THIẾT KẾ VI MẠCH TƯƠNG TỰ CHƯƠNG 2: CMOS Technology

Transcription:

Leading Edge Logic Comparison March 9, 2018 Scotten W. Jones President IC Knowledge LLC sjones@icknowledge.com

Logic Standard Cell Logic designs are created using standard cells. The cell height is the number of tracks multiplied by the metal pitch. Tracks and pitch are measured in metal 2. A 7.5 track cell is shown, one half of the power and ground rail heights are in the cell above and cell below respectively. 7.5 track logic cell 2

Metal 2 Pitch The figure on the right presents the metal 2 pitch for the four companies driving the lading edge for logic: GLOBALFOUNDRIES Intel Samsung TSMC Solid filled markers are actual values, unfilled markers are IC Knowledge estimates. The trend is based on actuals only. M2 Pitch (nm) 1,000 100 10 GF Act GF Est Intel Act Intel Est SS Act SS Est TSMC Act TSMC Est Trend 1 2000 2005 2010 2015 2020 2025 Year M2 Pitch Trend 3

Track Heights The figure on the right presents the track heights for the four companies driving the lading edge for logic. Solid filled markers are actual values, unfilled markers are IC Knowledge estimates. Reducing track height to drive shrinks is becoming more prevalent with recent and projected processes below the trend line. The trend is based on actuals only. Track Height 100.00 GF Act GF Est Intel Act Intel Est SS Act SS Est TSMC Act TSMC Est Trend 10.00 1.00 2000 2005 2010 2015 2020 2025 Year Track Heights 4

Cell Heights Combining the previous two slides we get the figure on the right presenting the cell heights for the four companies driving the leading edge for logic. Solid filled markers are actual values, unfilled markers are IC Knowledge estimates. The trend is based on actuals only. Cell Height (nm) 10,000 1,000 100 10 GF Act GF Est Intel Act Intel Est SS Act SS Est TSMC Act TSMC Est Trend 1 2000 2005 2010 2015 2020 2025 Year Cell Heights 5

Cell Widths The cell width is related to the contact poly pitch (CPP). The number of CPPs that make up the cell width depend on the cell type and whether the cell has a double diffusion break (DDB) or single diffusion break (SDB). A DDB adds an addition one half CPP to each side of a cell so that in the case illustrated on the right the cells are 3x CPP for the DDB or 2x CPP for the SDB. Double versus single diffusion break 6

SDB and DDB Usage For the four manufacturers: GLOBALFOUNDRIES Intel Samsung TSMC There are a variety of SDB and DDB usages depending on node and sometimes process variants within a node. 7

Contacted Poly Pitch The figure on the right presents the contacted poly pitch for the four companies driving the lading edge for logic. Solid filled markers are actual values, unfilled markers are IC Knowledge estimates. Note that CPP is limited to about 40nm due to device issues. The trend is based on actuals only. Contacted Poly Pitch (nm) 1,000 100 GF Act GF Est 10 Intel Act Intel Est SS Act SS Est TSMC Act TSMC Est Trend 1 2000 2005 2010 2015 2020 2025 Year Contacted Poly Pitch 8

Logic Cells Two Input NAND Cell One of the most common logic cells is the two input NAND gate. The figure at the right illustrates DDB and SDB two input NAND cells. A DDB - two input NAND cell is 4x CPP wide and a SDB - two input NAND cell is 3x CPP wide. Double versus single diffusion break Two input NAND cells 9

More Logic Cells Scanned Flip Flop Another common logic cell is the Scanned Flip Flop (SFF). The figure at the right illustrates DDB and SDB SFF. A DDB - SFF cell is 20x CPP wide and a SDB - SFF cell is 19x CPP wide. Double versus single diffusion break Scanned flip flop 10

2 Input NAND Cell Width Taking into account CPP and DDB versus SDB usage by company cell widths for two input NAND cells can be calculated. The plot on the right resents the cell widths for the four companies of interest by year. The trend is based on actuals only. NAND Cell Width (nm) 10,000 1,000 GF Act Intel Act SS Act TSMC Act Trend GF Est Intel Est SS Est TSMC Est 100 2000 2005 2010 2015 2020 2025 Year 2 Input NAND Cell Width 11

Scanned Flip Flop Cell Width Taking into account CPP and DDB versus SDB usage by company cell widths for scanned flip flop cells can be calculated. The plot on the right resents the cell widths for the four companies of interest by year. The trend is based on actuals only. Scanned Flip Flop Cell Width (nm) 10,000 1,000 GF Act Intel Act SS Act TSMC Act Trend GF Est Intel Est SS Est TSMC Est 100 2000 2005 2010 2015 2020 2025 Year Scanned Flip Flop Cell Width 12

Logic Process Density From the previous five slides it can be seen that the width of a logic cell depends on SDB versus DDB and what kind of cell it is. Intel has proposed revising an old metric of logic density based on a weighting of 60% two input NAND cells and 40% scanned flip flops. This meant to reflect the rough prevalence of the cell usage in logic designs. A cell sizes for two input NAND and scanned flip flops are the cell height and widths outlined previously. A two input NAND cell contains 4 transistors and a scanned flip flop cell contains 36 transistors. Based on the cells sizes and transistors per cell the cell type weighting can be used to produce transistor densities by process. 13

Transistor Density Using the proposed Intel transistor density metric, process density by year can be compared by year for the four companies of interest. The trend line is based on actuals only. Transistor density (MTx/mm 2 ) 1,000.00 100.00 10.00 GF Act GF Est Intel Act Intel Est SS Act SS Est TSMC Act TSMC Est Trend 1.00 2000 2005 2010 2015 2020 2025 Year Millions of Transistors Per Square Millimeter 14

Standard Node Based on Transistor Density The plot on the right plots the stated node for technologies versus transistor density for the four leading edge logic companies. Only actual values are plotted. Node 100 10 y = 183.95x -0.706 R² = 0.9233 1 0.00 20.00 40.00 60.00 80.00 100.00 120.00 Transistor density (MTx/mm2) Standard Node Versus Transistor Density 15

Standard Node Trend 100 GF Act GF Est Intel Act Intel Est SS Act SS Est TSMC Act TSMC Est 43.05 21.18 Standard Node (nm) 10 24.42 22.74 17.93 13.23 14.15 14.64 10.85 11.47 5.96 16.65 10.93 10.11 12.09 6.83 7.06 7.31 7.21 6.03 5.06 4.13 4.07 4.35 3.28 4.57 1 2008 2010 2012 2014 2016 2018 2020 2022 Year 16

Conclusion A detailed analysis of known and estimated nodes has been performed accounting for M2P, CPP, Tracks and DDB versus SDB. A standard node versus transistor density metric has been developed. The smallest standard node by year has been calculated and plotted for each of the four leading edge logic producers. Accounting for all factor intel has a slight standard node lead in 2017 but we expect TSMC to take the standard node lead in 2018. We expect TSMC and Samsung to maintain standard node leads over Intel from 2019 through 2022. 17