ic requirements for next generation systems
play

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

IC Requirements For Next Generation Systems Club Vivado Users Group Malcolm Penn Oct 2015 Chairman & CEO, Future Horizons Slide 1 Your Passport To Future Horizons Analysing The Facts The FUD, The Hype & Delusion Economy


  1. IC Requirements For Next Generation Systems Club Vivado Users Group – Malcolm Penn – Oct 2015 Chairman & CEO, Future Horizons Slide 1

  2. Your Passport To Future Horizons Analysing The Facts The FUD, The Hype & Delusion  Economy  Perception  Unit Demand  Emotion  Fab Capacity  Fashion  ASPs  Sentiment The Global Semiconductor Industry Analyst “Making Sense Of The Industry Tea Leaves” (Google “Future Horizons” or “Malcolm Penn Semiconductors” For More Details) Slide 2

  3. 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

  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 opinions, 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 4

  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 5

  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 6

  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 7

  8. Shame No-One Told Samsung & TSMC Memory Logic 2006 2010 2012 2013 2014 2015 2009 With 7nm, 5nm, 3nm, 2,25nm, 1.8nm & 1.3nm Well Understood (Whether anybody can afford them is another matter altogether!) Slide 8 Slide 8

  9. 3D NAND … Tough But Getting There Slide 9

  10. Supply’s Not An Issue Either (If Pre-Ordered*) Foundry Market By Feature Size “We Do Not Build Speculative Capacity” – Dr Morris Chang, Jan 2015* (FH Advisory … Net New Capacity Is A One-Year Lead Time Item) Slide 10

  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 11

  12. What Is Makimoto’s Wave?* * Named by D. Manners Standardization (Electronics Weekly , Jan. 30,1991) Field Memories Program- Standard Micro- mability Discretes processors '67 '87 '97 '77 '07 '57 Custom LSIs for TVs, ASICs Calculators Standardized in Manufacturing but Customized in Application Customization Source: Dr. Tsugio Makimoto Slide 12

  13. Semiconductor Pendulum (Custom vs Standard Enigma) Customization Standardization Source: Dr. Tsugio Makimoto Slide 13

  14. Makimoto’s Wave Extension Standardization Highly Flexible Super Integration Field MPU & Standard Program- HFSI Memory '67 '77 '87 '97 '07 '17 Discrete mability ‘ 2 7 '57 Custom SoC/SiP ASIC LSI Customization ● 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 Source: Dr. Tsugio Makimoto Slide 14

  15. Process Evolution Remains Key A9 A9X* *14nm A9 (Samsung) 96mm 2 / 16nm A9X (TSMC) 105mm 2 Slide 15 Slide 15

  16. Except For … The Interconnect! 14/16nm >20 Million Transistors On A Pin Head  SRAM Cell ~ 0.070µ 2 – TSMC / 0.059µ 2 - Intel 12-Layer Metal, Top Layer 0.1mm Wide (L1 Too Wide Also)  For Many Chips Shrinks Below14/16nm Are ‘Irrelevant’ (No Gain) 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) Slide 16

  17. Spacew ire Network On A Chip Source: Dundee University/Future Horizons Slide 17

  18. (MEMS) Sensors Key Too  Smartphone  Automotive – Engine Management  Automotive – Chassis & Safety  Printers  Home Automation  Industrial  Projection Displays 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’) Source: Future Horizons (2015 MEMS Market Update Report) Slide 18

  19. MEMS Product Examples 10-Axis Gyroscope Microphone Pressure Sensor Source: Freescale Source: Knowles/ChipWorks Source: OmronD InkJet Print Head RF Switching Camera Focusing Source: STM Source: IOP Source: InvenSense Source: Future Horizons (2015 MEMS Market Update Report) Slide 19

  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 Slide 20

  21. Chip Market Drivers Slide 21

  22. Smartphone – Ubiquitous & Transformative From Jan 2007 Launch To ‘Planet Of The Phones’ To Phonosapiens … 21 st Century Oods? From Homosapiens Slide 22 Slide 22

  23. From Automation To Design Services Slide 23 Slide 23

  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 Slide 24

  25. Micro-Fluidics (Lab-On-A-Chip) Source: Berkeley Source: ElveFlow  Similar Techniques To Inkjet-printer, Already Surprisingly Large Market (~$2b)  Highly Specialised BUT Chemistry Dominates This Field NOT The MEMS Device 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

  26. Data Explosion 2015 = 2x (only) Source: Intel Slide 26 Slide 26

  27. Semiconductor Innovation So ‘No Pressure’ There Folks … !! Slide 27 Slide 27

Recommend


More recommend