Semiconductor Trends 2005
180
160 s
t 140
120 Uni 100 of 80 ons
i 60 l l i 40 M 20
0 Windows Servers SuperCs Nick Tredennick, Editor Gilder Technology Report [email protected] 21 September 2005 Nick Tredennick 1 Overview • Major trends affecting the microprocessor market – Value PC – Value transistor – Emerging economies • Microprocessors – Computer microprocessors – Embedded microprocessors – Configurable microprocessors – PLD microprocessors
21 September 2005 Nick Tredennick 2 Moore’s Law
21 September 2005 Nick Tredennick 3 Moore’s Law: Wafer Yield
• 37 die/wafer • 177 die/wafer • 28 good die • 167 good die • Yield: 75% • Yield: 94% Halving dimensions yields 6x good die! 21 September 2005 Nick Tredennick 4 Top View: Field-Effect Transistor
21 September 2005 Nick Tredennick 5 The Microprocessor • 10 years of Moore’s-law progress led to the microprocessor • The second generic component • Raised engineers’ productivity • Problem-solving became programming • Grew to billions of units/year • Stalled progress in design methods for thirty years
21 September 2005 Nick Tredennick 6 The Personal Computer • 10 years of microprocessor progress led to the PC • Dominated the industry for 20 years • Supply of performance grows with Moore’s law • Demand grows more slowly • Diverging growth in supply and demand leads to the value PC
21 September 2005 Nick Tredennick 7 The PC Is Good Enough
21 September 2005 Nick Tredennick 8 The Path To The Value Transistor
21 September 2005 Nick Tredennick 9 Electronic Systems Market Segments
21 September 2005 Nick Tredennick 10 Computer Markets
180
160 s
t 140
120 Uni 100 of 80 ons
i 60 l l 40 Mi 20
0 Windows Servers SuperCs
21 September 2005 Nick Tredennick 11 Microprocessor Markets
8000
7000 s t i
n 6000
U 5000 f
o 4000 s n
o 3000 i l l 2000 Mi 1000
0 Embedded PCs Servers
21 September 2005 Nick Tredennick 12 The Cost-Performance Segment
21 September 2005 Nick Tredennick 13 Cost-Performance-Per-Watt
21 September 2005 Nick Tredennick 14 Microprocessors Are Unsuitable
21 September 2005 Nick Tredennick 15 Programmers And Logic Designers
• Programmers optimize software – Languages – OS – Compilers The Users Manual – Applications is the Programmers (problematic) bridge • Logic designers optimize hardware – Microprocessors – Memory Logic designers 21 September 2005 Nick Tredennick 16 Scenic View
21 September 2005 Nick Tredennick 17 Scenic View in Fog
21 September 2005 Nick Tredennick 18 Scenic View with Lens
21 September 2005 Nick Tredennick 19 Scenic View and Photoshop
21 September 2005 Nick Tredennick 20 Microprocessors and ASICs
Microprocessor
• For the ultimate in flexibility, programmers map the application onto a general-purpose microprocessor. Programmers
• For the ultimate in performance, logic Application designers map the application into a custom circuit.
ASIC Logic designers 21 September 2005 Nick Tredennick 21 ASICs & Microprocessors
21 September 2005 Nick Tredennick 22 Microprocessor Evolution
Dynamically reconfigurable microprocessor Run-time reconfigurable microprocessor Microprocessor Stretch Design-time Programmers configurable microprocessor ARC MIPS Tensilica
ASIC
FPGA Logic designers 21 September 2005 Nick Tredennick 23 Design-Time Configurable Microprocessor
Most of the application runs as execution of general-purpose Profile the instructions application
Programmers
Create custom Design-time hardware and configurable Application instructions to microprocessor accelerate critical application sections
Logic designers 21 September 2005 Nick Tredennick 24 Design-Time Configurable Microprocessor
• Profile the application • Create custom instructions for critical code sections • Build specialized execution units • Can be 10 to100 times faster than a general-purpose microprocessor on the target algorithm • Examples: ARC and Tensilica • Customized microprocessor limitations – Requires logic designers – Creates an application-specific, limited-function microprocessor – Accelerates only critical sections
21 September 2005 Nick Tredennick 25 Run-Time Reconfigurable Microprocessor
Most of the application runs as execution of Profile the general-purpose application instructions
Run-time Create custom reconfigurable instructions in an microprocessor FPGA fabric to accelerate critical application sections Programmers Application
Logic designers 21 September 2005 Nick Tredennick 26 Run-Time Reconfigurable Microprocessor
• Build a general-purpose microprocessor with integrated FPGA fabric • Profile the application • Create custom instructions for critical code sections • Build custom execution units in FPGA fabric • Can be 10 to 100 times faster than a general-purpose microprocessor on the target application • Example: Stretch • Run-time reconfigurable microprocessor limitations – Accelerates only statically identifiable critical sections – Limited to problems for which profiling works – Profiling is difficult
21 September 2005 Nick Tredennick 27 ASICs & Microprocessors
21 September 2005 Nick Tredennick 28 Dynamically Reconfigurable Microprocessor
Dynamically Create custom reconfigurable instructions in a microprocessor custom fabric to accelerate the entire application Programmers Application
Logic designers 21 September 2005 Nick Tredennick 29 Dynamically Reconfigurable Microprocessor • Each cycle creates a new microprocessor implementation – Each cycle creates a custom circuit (Ascenium instruction) representing hundreds to thousands of conventional instructions • Programmed using ANSI-standard programming languages (e.g., C/C++) • Tens to 100s of times faster than a general-purpose microprocessor • Dynamically reconfigurable microprocessor limitations – There are none on the market today • Until Ascenium, no one has figured out how to “program” a dynamically reconfigurable circuit – VCs don’t understand it
21 September 2005 Nick Tredennick 30 Reconfigurable Systems Emerge
21 September 2005 Nick Tredennick 31 Situation
What Value Who • PLDs $3B logic designers • ASICs $30B logic designers • Microprocessors $40B programmers
PLDs and microprocessors are usurping a declining ASIC market. Microprocessors (and their derivatives) will win.
21 September 2005 Nick Tredennick 32 Supply and Demand: ASICs & PLDs
21 September 2005 Nick Tredennick 33 Battles are fought at the leading edge; the war goes to what is good enough. • RISC vs. CISC – Proving you can build a career on smoke and mirrors • Assembler vs. High-level language • Workstations vs. PCs – Build for volume and performance follows; build for performance and languish • ASICs vs. PLDs
21 September 2005 Nick Tredennick 34 Semiconductor Development Fixed Resources and The Value PC Fixed Algorithms tion nta Fixed Resources me ple Im ing Variable Algorithms sed mm -ba gra nt ion ro icie uct s p eff tr me , in rs o Value PC Ins ec ate ure g: b equ act vin ad uf sol le, an l m dab s, m tal ble for ool r s ro f , t ea 2003 P s: a ips -y ion ch : 30 lut ed: ods Circuit-based Implementation o S nch th tre n Me s En sig me De olu te v ina nt om me s d lop w tem eve s la sys p d re’ Microprocessor i o Computer d h o ere te c M eth ina on T om nds s d epe PC ss d 1971 re 194x g Pro Cost Performance 21 September 2005 Nick Tredennick 35 The End of Design’s cy ien ic ff E Fixed Resources 30-Year Stall er w Fixed Algorithms Po tion nta s Fixed Resources me m s e te ip mpl ng s h Variable Algorithms d I mi sy c se m e d -ba gra ent il e ion pro fici b d uct es ef o ed tr m , in rs M b o e Value PC Ins ec at ure m g: b equ act E vin ad uf sol le, an l m dab s, m tal ble for ool r s ro f , t ea 2003 P s: a ips -y ion ch : 30 lut ed: ods Variable Resources Circuit-based Implementation o S nch th tre n Me s Variable Algorithms En sig me De olu te v ina nt Generic chips om me s d lop w New non-volatile tem eve s la sys p d re’ Microprocessor i o memory, new PLDs Computer d h o ere te c M eth ina on Instruction-based T om nds s d epe processing is PC ss d 1971 re 194x g inadequate Pro Cost Performance Cost Performance 21 September 2005 Nick Tredennick Per Watt 36 Consequences • Rise of mobile applications – New non-volatile memories • Rise of foundries – Rise of soft (IP) cores – Horizontal fragmentation of integrated device manufacturers • Rise of non-volatile PLDs • Rise of reconfigurable systems • Growing market for embedded microprocessors – Tethered: traditional role – Mobile: supervisory role
21 September 2005 Nick Tredennick 37 21 September 2005 Nick Tredennick 38 Microprocessors • x86 AMD, Intel, Via •ARC ARC •ARM ARM • MicroBlaze Xilinx • MIPS MIPS • Nios Altera • PowerPC IBM, Freescale • SPARC Sun • Tensilica Stretch, Tensilica • Old stuff Everyone
21 September 2005 Nick Tredennick 39 Microprocessor Applications
• Supercomputers • Workstations and servers •PCs • Embedded systems – Automobiles – Cameras – Cell phones – Game players – MP3 players – Set-top boxes
21 September 2005 Nick Tredennick 40 Computer Microprocessors •x86 – AMD – Intel – Via • Proprietary – IBM – Freescale – Sun
21 September 2005 Nick Tredennick 41 Embedded Microprocessors • Microprocessor advantages – Flexibility – High-volume production – Usable by programmers • Microprocessor limitations – Too slow – Too much power
21 September 2005 Nick Tredennick 42 Embedded Microprocessors • x86 AMD, Intel, Via •ARM ARM • PowerPC IBM, Freescale • Old stuff Everyone – Triscend (Xilinx)
21 September 2005 Nick Tredennick 43 Configurable Microprocessors •ARC ARC • Ascenium Ascenium • MIPS MIPS • Nios Altera • Tensilica Stretch, Tensilica
21 September 2005 Nick Tredennick 44 PLD Microprocessors •Altera – Nios (soft) • Xilinx – MicroBlaze (soft) – PicoBlaze (soft) – PowerPC (hard)
21 September 2005 Nick Tredennick 45