IC Requirements For Next Generation Systems Club Vivado Users Group - - PowerPoint PPT Presentation

Slide 1

IC Requirements For Next Generation Systems

Club Vivado Users Group – Malcolm Penn – Oct 2015 Chairman & CEO, Future Horizons

Your Passport To Future Horizons

The Global Semiconductor Industry Analyst

“Making Sense Of The Industry Tea Leaves”

(Google “Future Horizons” or “Malcolm Penn Semiconductors” For More Details) Analysing The Facts

 Economy  Unit Demand  Fab Capacity  ASPs

The FUD, The Hype & Delusion

 Perception  Emotion  Fashion  Sentiment

Slide 2

Chronology & Background

 1989 – Company Founded (Apr 1)  Worldwide Semiconductor Industry Focus  European & Russia/CIS Semiconductor Industry Specialisation  Full Spectrum – Advanced Research To Market Development  Market Research, Industry Analysis & Training  Custom Consulting & Marketing Studies  Due Diligence & IP/Product Positioning  Business Development Opportunities & 1-2-1 Contacts  Start-Ups Through Large Corporations & NXD Support

• Off-The-Shelf Research Reports & Industry Intelligence/Analysis
• Bespoke Research Assignments - From Proof Of Concept To Full Market Development
• Due Diligence Analysis - From Seed Funding Through IPO
• Competitive Benchmarking & Positioning - From the Basic IC Design & Technology Up
• Unique Combination Of Chip Design Know-How With Market & Business Insight

~5 Decades Of Semiconductor Industry Experience

Longer Than ANY Other Industry Analyst & Most Industry Execs

(Google “Malcolm Penn Drums” For My Near Alternative & Formulative Early Career!)

Slide 3

Slide 4

Ten Industry Mega-Challenges*

 New Start-Up Famine: Due to the high costs and loss of VC appetite driving

Shift to IP

 Fabless Market Saturation: With all IDMs now fabless the fabless sector can

no longer ‘outperform’ the market … it is the market

 Foundry Supply: With TSMC dominant, they can now only grow ‘with the market’  Fab-Tight Supply: Net new capacity now built to order not expectation  Virtual OEM: The emergence of firms such as Amazon, Apple and maybe Google  Market Opportunities: Need for more substantial research vs superficial

• pinions, blogs and over-hyped head-line grabbing articles (e.g. the IoT Fiasco)

 Technology Challenges: Every new node (and transistor design) here on out

will be revolutionary not evolutionary

 Industry Consolidation: Reducing the overall market pie for the down-stream

providers and supplier choice for the up-stream customers

 ‘More Than Moore’: Assuming deeper importance as systems become smarter,

more intelligent, interconnected and communicative

 New Design Techniques: Addressing the increasing occurrence of errors in

the logic execution

* FH Research Report (Feb 2015) Subscribe To Our Regular Update Reports & Industry Briefings

Slide 5

Three Patterns Of Semiconductor Innovation

 Disruptive Innovation

• Invention Of Transistor (Shockley, Bardeen, Brattain: 1947)
• Invention Of The IC (Kilby: 1958 / Noyce: 1959)
• Microprocessor Development (Faggin, Shima, Hoff, Mazor: 1971)

 Exponential Innovation

• Moore’s Law (Gordon Moore: 1965)

 Cyclical Innovation

• Makimoto’s Wave (Tsugio Makimoto*: 1991)

* Previously GM of Hitachi SC (1959-2001); Spearhead of 6147 High-Speed CMOS Intel 2147 SRAM Replacement

Slide 6

Three Patterns Of Semiconductor Innovation

 Disruptive Innovation

• Invention Of Transistor (Shockley, Bardeen, Brattain: 1947)
• Invention Of The IC (Kilby: 1958 / Noyce: 1959)
• Microprocessor Development (Faggin, Shima, Hoff, Mazor: 1971)

 What Next?

• Post-CMOS Scaling (Materials & Structures)
• Biological ICs/Systems (Grown Inside The Final Package)
• Quantum Computing (When We Finally Figure Out How)

Slide 7

Three Patterns Of Semiconductor Innovation

 Exponential Innovation

• Moore’s Law (Gordon Moore: 1965)

 What Next?

• Moore’s Law Is Dead (It’s Over At 28nm, That Much We’ve Been Told!!)

Slide 8

Shame No-One Told Samsung & TSMC

Logic Memory With 7nm, 5nm, 3nm, 2,25nm, 1.8nm & 1.3nm Well Understood

(Whether anybody can afford them is another matter altogether!)

2006 2009 2010 2012 2013 2014 2015 Slide 8

3D NAND … Tough But Getting There

Slide 9

Supply’s Not An Issue Either (If Pre-Ordered*)

Foundry Market By Feature Size

Slide 10

“We Do Not Build Speculative Capacity” – Dr Morris Chang, Jan 2015* (FH Advisory … Net New Capacity Is A One-Year Lead Time Item)

Slide 11

Three Patterns Of Semiconductor Innovation

 Cyclical Innovation

• Makimoto’s Wave (Tsugio Makimoto: 1991)

 What Next?

• Now This Question IS Interesting! (Not Just For Xilinx Either )

Slide 12

What Is Makimoto’s Wave?*

Standard Discretes Custom LSIs for TVs, Calculators Memories Micro- processors ASICs Field Program- mability

Standardization Customization

'67 '77 '87 '97 '57 '07

Standardized in Manufacturing but Customized in Application ＊ Named by D. Manners (Electronics Weekly , Jan. 30,1991)

Source: Dr. Tsugio Makimoto

Slide 13

Semiconductor Pendulum (Custom vs Standard Enigma)

Standardization Customization

Source: Dr. Tsugio Makimoto

Slide 14

Makimoto’s Wave Extension

'17

SoC/SiP

‘２7

Highly Flexible Super Integration HFSI

• Why HFSI? Same Reason As Before …

Increasing design cost, Fragmented market

• HFSI Technological Breakthroughs

1) Integration of multi-functions incl. FPGA 2) Emergence of high-performance NVRAM

Standard Discrete Custom LSI MPU & Memory ASIC Field Program- mability

Standardization Customization

'67 '77 '87 '97 '57 '07

Source: Dr. Tsugio Makimoto

Slide 15

Process Evolution Remains Key

*14nm A9 (Samsung) 96mm2 / 16nm A9X (TSMC) 105mm2

A9 A9X* Slide 15

Except For … The Interconnect!

14/16nm >20 Million Transistors On A Pin Head

• SRAM Cell ~ 0.070µ2 – TSMC / 0.059µ2 - Intel

Chip Size Dominated By Interconnect Not Gates

• Probably Why Some Firms Skipping 10nm For 7nm

Needs New Interconnect Techniques

• On-Chip Wireless vs Track? (1cm Range / Tens of GHz, Not ISM Bands)
• On-Chip Optical? (Alternative to Wireless)
• Network On Chip? (e.g. Dundee Spacewire Technology)

12-Layer Metal, Top Layer 0.1mm Wide (L1 Too Wide Also)

• For Many Chips Shrinks Below14/16nm Are ‘Irrelevant’ (No Gain)

Slide 16

Spacew ire Network On A Chip

Source: Dundee University/Future Horizons

Slide 17

 Smartphone  Automotive – Engine Management  Automotive – Chassis & Safety  Printers  Home Automation  Industrial  Projection Displays

(MEMS) Sensors Key Too

Source: Future Horizons (2015 MEMS Market Update Report)

Slide 18

Plus

• Low-Volume (High-Value ) Aerospace & Defence

(enabling spin-off products such as commercial ‘drones’)

• ‘Wearables’’, Medical Products & Other ‘IoT’ Applications
• Gaming, Robotics & Toys (‘Other Sales’)

MEMS Product Examples

Microphone

Source: Knowles/ChipWorks

InkJet Print Head RF Switching

Source: IOP Source: STM Source: InvenSense

Camera Focusing

Source: Freescale

10-Axis Gyroscope Pressure Sensor

Source: OmronD Source: Future Horizons (2015 MEMS Market Update Report)

Slide 19

Slide 20

Other Key Challenges

Connectivity: Currently – Low Number Of High Bandwidth, Single Connections IoT – ‘Trillion’ Low Bandwidth, Unique IP Addresses (Router vs IPv6 Issue) Security: Currently – Firewall IoT – Built Into Device (From Thermostats To Dolls; From TVs To Cars) Processor: Currently – S’Ware Hackable MPU (Intel, PowerPC, ARM, XMOS …) IoT – Reverse-Engineer ‘Impenetrable’ FPGA Code Software: Currently – 1-10 Threads, Serial Code – Point-To-Point Network Centric IoT – Millions Of Threads, Parallel Code – Neural Network Centric SoC Design: Currently – Independent Design Teams, Chip Partitioning-Based IoT – Agile Development, Collaborative, Cross-Functional, 2-Week Sprints

Chip Market Drivers

Slide 21

Slide 22

Smartphone – Ubiquitous & Transformative

From Jan 2007 Launch

From Homosapiens

To ‘Planet Of The Phones’

To Phonosapiens … 21st Century Oods?

Slide 22

Slide 23

From Automation To Design Services

Slide 23

Slide 24

To The Much Over-Hyped ‘IoT’

Industry & Governments Talk About IoT As If It Were One Space, One Solution – It Is Of Course Not … It Represents A Wide Range Of Markets, Applications, Technologies & (Eventually Real) Opportunities

Slide 24

Micro-Fluidics (Lab-On-A-Chip)

 Similar Techniques To Inkjet-printer, Already Surprisingly Large Market (~$2b)  Highly Specialised BUT Chemistry Dominates This Field NOT The MEMS Device

Source: ElveFlow Source: Berkeley

Think Vitamins & ‘Voodoo’ Healthcare Sales … This Market Is Huge

(But Not For The Chip Suppliers… Think Apple & Amazon!!)

Source: Future Horizons (2015 MEMS Market Update Report)

Slide 25

Slide 26

Data Explosion

Source: Intel

Slide 26

2015 = 2x (only)

Slide 27

Semiconductor Innovation

So ‘No Pressure’ There Folks … !!

Slide 27

Slide 28

Makimoto’s Wave – Observation Standardization Customization

'17

SoC & SiP

‘27

HFSI Standard Discrete Custom LSI MPU & Memory ASIC Field Program- mability

'67 '77 '87 '97 '57 '07 SASC?

Amplitude Of Sine Wave Decreasing At Successive Periods Each Less Flexible / Less Programmable Than Before …

Source: Dr. Tsugio Makimoto / Future Horizons

‘47 ‘37

SASC ? ?

Slide 29

Makimoto’s Wave – Ultimate End Game

Self Aware Self Configuring ICs (Systems)

 Think For Themselves & Configure Accordingly  Intelligent ICs … The Ultimate Hardware Solution

Be Careful What You Wish For!

Slide 30

Contact Details – w w w.futurehorizons.com

Future Horizons Ltd 44 Bethel Road Sevenoaks, Kent TN13 3UE, England T: +44 (0)1732 740440 F: +44 (0)1732 464270 E: mail@ futurehorizons.com Regional Offices In The UK & Russia Affiliates In Europe, India, Israel, Japan, Russia & USA Follow us on: Tw itter Face Book LinkedIn

Thank You

Please E-Mail/Contact Future Horizons For A Copy Of This Presentation