A START-UP S PROSPECTIVE TO TECHNOLOGY CHOICE AND IC DEVELOPMENT IN DEEP SUBMICRON CMOS

Transcription

1 A START-UP S PROSPECTIVE TO TECHNOLOGY CHOICE AND IC DEVELOPMENT IN DEEP SUBMICRON CMOS JED Hurwitz CTO & VP Engineering, Gigle Semiconductor 14 May 2008

2 CONTENTS OF TALK Who am I Aspects of a successful technology start-up Types of Start-Up Stages of Start-Up What is the Start Up s Perspective to Technology Choice Enabling the Secret Sauce Risk, Cost, Access Building Value & Keeping Competitiveness Deep Submicron IC Development Simplified view of any CMOS technology roadmap Trends in deep submicron Major Risks & Challenges Case Study Who is Gigle Semiconductor What is our market Our Approach Our Development Flow Summary

3 WHO AM I JED Hurwitz 21 years in Mixed Signal Semiconductors - Plessey, Matra MHS, Vision Group (sold to ST Microelectronics), Consultant Expertise - Focused on Analog + System solutions - Always been close to the technology & the market - Technically led ST s mobile imaging activity to 10 s Millions of chips, 100 s M$ of product Co-founded Gigle Semiconductor in Oct Fabless semiconductor company in the Home Networking space - Targeting a new generation of Anywire SoC

4 CONTENTS OF TALK Who am I Aspects of a successful technology start-up Types of Start-Up Stages of Start-Up What is the Start Up s Perspective to Technology Choice Enabling the Secret Sauce Risk, Cost, Access Building Value & Keeping Competitiveness Deep Submicron IC Development Simplified view of any CMOS technology roadmap Trends in deep submicron Major Risks & Challenges Case Study Who is Gigle Semiconductor What is our market Our Approach Our Development Flow Summary

5 Aspects of a successful Start-up 1 Target a suitable Market Should be sustainable and ripe for expansion, either through growth or change. Build a strong team There is nothing more valuable (invest-able) than people who have done it before. Keep that team healthy. Make sure there is some Secret Sauce IP, Know-how, Understanding there has to be something that you re doing that is different to everyone else. Choose Good Partners Financial, Technology, Manufacturing, Customer. Build the Ecosystem.

6 Aspects of a successful Start-up 2 Plan to build a sustainable advantage The most difficult question to answer is why can t someone else do it? Use your money wisely Build value, and ensure there are clear inflection points for raising finance Keep Focus This IS one of your underlying advantage DON T lose it! Timing There are no prizes for arriving too late OR sometimes too early Know your markets and competitors Need to be competitive when you arrive

7 Types of Technology Start-Up The execution play Market Certainty Linked with scale of play The innovation play Risk? Cost? Time? Return? The disruptive play Competitive Advantage Different types of play you re going to have to justify :- Is the market ready? How are you going to make a difference? Why can t the existing players do what you re doing? Technology choice should match the plan Needs to enable the competitive advantage and match the scale of the play

8 Stages of Technology Start-Up Initial Conception and Funding Market & technology analysis Gaining Credibility in all areas Finding partners Development Building the Team, Technology, Market and Product Execution, execution, execution Product launch Getting Customers Delivering to Customers Growth Revenue Developing product derivatives and diversification Continuing technology development Risk Cost, Time

9 CONTENTS OF TALK Who am I Aspects of a successful technology start-up Types of Start-Up Stages of Start-Up What is the Start Up s Perspective to Technology Choice Enabling the Secret Sauce Risk, Cost, Access Building Value & Keeping Competitiveness Deep Submicron IC Development Simplified view of any CMOS technology roadmap Trends in deep submicron Major Risks & Challenges Case Study Who is Gigle Semiconductor What is our market Our Approach Our Development Flow Summary

10 Enabling the Secret Sauce Every Start-Up is different, but normally there is one key part ( secret sauce ) that needs enabling to ensure the plan can be executed. Could be a particular device Could be a particular circuit or algorithm Could be a particular system combination Could be a level of integration Could be a level of performance Focus on the technology that makes the Secret Sauce work Deep Submicron strengths Unit cost of gates & memory, speed, power, technology diversity Deep Submicron challenges NRE, timescale, DFM, IP availability, power, technology diversity

11 Risk, Cost, Access Deep submicron IC development is risky, Think of all the things that need to work - Only as good as your last mistake Bleeding edge technology has greater risks - Technology and Development Flow still maturing Start-ups have additional risks - Building a good team quickly - Gaining credibility and technology access - Finance Manage the risk Don t take on everything, BUT keep in-house the value. - Look for proven partners - Subcontract Put in place good processes. - Perform regular risk analysis and project reviews - Include key milestones in the plan Know the risks of your Architecture & concentrate on the risks that will effect your market entry Use the best technology that you can manage the risk

12 Risk, Cost, Access Money as a Start-Up in development is a precious commodity The smaller the geometry the greater the development cost! - Required CAD Tools - Premium to get silicon proven IP - Mask Sets - Cycle Times The smaller the geometry the lower the unit cost, and the greater the barrier for entry for competitors Use the cheapest technology that can enable your secret sauce Not all technology is available when and how you need it, Need to earn the right to access leading technology at foundry Not all technology has the maturity needed at the appropriate stage of the project IP generally becomes available when a technology matures Use the best technology that you can access

13 Building value & keeping competitiveness Not everyone can develop on deep submicron. This can be part of the value in the company. Certain skills have a premium - Analog, RF - Embedded FW and CPU architecture Demonstrating Right First Time methodology gains credibility Technology choice raises the barrier for entry Invest in the best technology that gives long term value If there is a market, your competitors will be moving technology too. What technology are they on now What technology will they be on when you come to market Are they able to add any of their own secret sauce Use at least as good a technology as your competitors will be when you come to market

14 CONTENTS OF TALK Who am I Aspects of a successful technology start-up Types of Start-Up Stages of Start-Up What is the Start Up s Perspective to Technology Choice Enabling the Secret Sauce Risk, Cost, Access Building Value & Keeping Competitiveness Deep Submicron IC Development Any CMOS technology roadmap Trends in deep submicron Major Risks & Challenges Case Study Who is Gigle Semiconductor What is our market Our Approach Our Development Flow Summary

15 Any CMOS Technology Roadmap Most technology roadmaps show at least one of these generic characteristics versus time whether it is for Unit Cost NRE Gate/memory density Active Power Leakage Power Speed Cycle time IP availability Choose the best technology that meets your requirements at both the start and the end of you project

16 Trends in deep submicron As we go from 90nm 65nm 45nm Transistors are getting smaller, designs need to be more complex to justify investment. Embedded memory becoming important, as technology crosses full system integration requirement thresholds The LP process is beginning to lead the GP process through the importance of mobile, wireless and power aware systems The transistor speed is not improving significantly for each new LP process, to contain the leakage issue. The transistor speed is improving significantly for each new GP process, but the leakage needs to be managed by design. Availability (and need) of >3.3V IO reducing Design flows are becoming more complex, crosstalk, DFM and statistical analysis becoming standard part of the development Mask costs are increasing but cheaper prototyping routes are being developed, so that full masks need only be used on products

17 Major Risks & Challenges Major Risk - achieving Right 1 st Time production silicon in realistic timeframes and budgets The economics of deep submicron, force major system integration, this by nature increases risk. - Complexity - Probability that one thing will be the weakest link Key risk management areas - IP/Block validation - Functional validation - Architectural flexibility - Silicon prove Hard functions - Pipe-clean the development Major Challenge - Design flow Managing the requirements Managing the design complexity and revisions Managing Power Managing Hard IP (Analog, RF, ) Managing validation Managing the backend Managing Design for Manufacturing

18 CONTENTS OF TALK Who am I Aspects of a successful technology start-up Types of Start-Up Stages of Start-Up What is the Start Up s Perspective to Technology Choice Enabling the Secret Sauce Risk, Cost, Access Building Value & Keeping Competitiveness Deep Submicron IC Development Simplified view of any CMOS technology roadmap Trends in deep submicron Major Risks & Challenges Case Study Who is Gigle Semiconductor What is our market Our Approach Our Development Flow Summary

19 WHO ARE WE - Gigle Semiconductor Fabless semiconductor company Founded in Oct 2005 Offices in Redwood City, Barcelona and Edinburgh 50 employees developing System-On-a-Chip (SOC) solutions 1st Generation: The GGL541 dual-channel, mesh solution for multimedia home networking over Anywire. to enable cost-effective, multimedia home networking. Greatest throughput and coverage Highest reliability Greatest implementation flexibility Lowest cost

20 Gigle s Market - The Multimedia Home Network By 2011 (worldwide): Media PC Digital Media Adapter Multi-Room Digital Video Recorder Game Console 733M network enabled devices 146M devices connected by powerline 87M devices connected by coax Home Gateway High Definition Entertainment Center Set-Top Box Source: isuppli, April 2007 Hi Def TV Hi Def DVD Player Home Network Attached Storage Requirements to address the Consumer Experience: Performance and coverage - delivers low latency and video without jitter or distortion Reliability - works everywhere, every time Quick and easy set-up - enables self-installation and reinstallation Low cost - is inexpensive enough to be embedded in all intelligent devices

21 A New Architectural Approach: Gigle s GGL541 Dual-Channel SOC Ad Hoc Mesh Networking to Increase Signal Range and Network Coverage 200 Mbps PHY throughput over powerline Fully compliant with Homeplug AV specifications Utilizes the Existing Wiring in the Home Small, low power System-On-Chip for lowest system cost GGL541 Cat 5 MII & GMII 10/100/1000 DFE MII & GMII 10/100/1000 MAC HomeplugAV MAC HomeplugAV AFE & DFE xtendnet TM CPU Memory mediaxtream TM MAC mediaxtream TM AFE & DFE OR OR Optimized and self-configured ad hoc mesh networking with built in repeater support Smart autoconfiguration QOS agent to extend coverage Enables 100+Mbps whole home coverage High reliability protocol for HDTV streams Encryption key exchange protocol IGMP snooping for IPTV multicast DLNA management middleware 1 Gbps PHY throughput over any wire Fully compliant with worldwide EMC regulations Dual-Channel for High Throughput Redundancy, Reliability, and Implementation Flexibility

22 Our Challenge To integrate 2 modems with 5X the performance for less cost than our competitors will integrate 1 modem when we arrive to market. Integrate 2 full AFEs (not yet achieved by our competitors) CPU centric approach, share resources Maximize clock frequency - Use the fastest process available - Process chosen for logic & memory - Analog designers forced to design on what they re given. Getting it right first time, shortening the time to market Choice of foundry, CAD, Subcontractors & IP Building a good team of people Choosing appropriate Technology proof points, De-risking the project through appropriate use of test chips It s NOT all about silicon.

23 Our Technology Choice vs Project Timeline What technology did the Secret Sauce need Good speed Good gate density Good Memory density 3v3 IO + Thick Metal Good partners (technology access, PDK s, IP) Project Start - Oct 2005 Choice 90nm Generic Just Mad Enough Competitors on 180nm or 130nm, with less integration, need to intercept in 2008 Most common question from VC s during funding can you do analog on 90nm, why don t you use 130nm Mid Project Dec 2006 Functionality and specification increased, die size increased, availability of half-node New question - should we move to 80nm, NO keep focus - IP already sourced, or in development, market waiting Near-End project 2008 Design progressed well, competitors have moved technology, but not architecture, competitive advantage is there and customers are waiting Most common question from VC s when can you do the next generation ;-)

24 Summary Deep submicron IC Development isn t easy or cheap, but it is manageable and can be rewarding SO be Just Mad Enough Choose your technology for the right reasons Focus on what delivers value Execute as clinically as you can Be paranoid, until its over Thank You