Introduction Mohammad Sharifkhani. The IC Growth CycleDensityPerformance Switching energy Cost per...

Introduction

Mohammad Sharifkhani

The IC Growth Cycle

Density Density

PerformancePerformance

Switching energySwitching energy

Cost per functionCost per function

Applications Applications

DevicesDevices

UsersUsers

RevenueRevenue

TechnologyTechnologyadvancesadvances

ConsumptionConsumptiongrowsgrows

Money!

Chen, ISSCC’07

Market share

Logic is there all the time

““Reduced cost is one of the big Reduced cost is one of the big attractions of integrated attractions of integrated electronics, and the cost electronics, and the cost advantage continues to increase advantage continues to increase as the technology evolves as the technology evolves toward the production of larger toward the production of larger and larger circuit functions on a and larger circuit functions on a single semiconductor substrate.”single semiconductor substrate.”Electronics, Volume 38, Electronics, Volume 38, Number 8, April 19, 1965Number 8, April 19, 1965

19601960 19651965 19701970 19751975 19801980 19851985 19901990 19951995 20002000 20052005 20102010

TransistorsTransistorsPer DiePer Die

101088

101077

101066

101055

101044

101033

101022

101011

101000

101099

10101010

Source: IntelSource: Intel

Moore’s Law - 1965

1965 Data (Moore)1965 Data (Moore)

Moore’s Law - 2005

40044004

80808080 808680868028680286

386™ Processor386™ Processor486™ Processor486™ Processor

PentiumPentium® ® ProcessorProcessorPentiumPentium® ® II ProcessorII Processor

PentiumPentium®® III Processor III ProcessorPentiumPentium®® 4 Processor 4 Processor

ItaniumItanium™ ™ ProcessorProcessor

TransistorsTransistorsPer DiePer Die

101088

101077

101066

101055

101044

101033

101022

101011

101000

101099

10101010

80088008

ItaniumItanium™ ™ 22 ProcessorProcessor

1K1K4K4K

64K64K256K256K

1M1M

16M16M4M4M

64M64M

256M256M512M512M

1G1G 2G2G

128M128M

16K16K

1965 Data (Moore)1965 Data (Moore)

MicroprocessorMicroprocessor

MemoryMemory

19601960 19651965 19701970 19751975 19801980 19851985 19901990 19951995 20002000 20052005 20102010


Silicon Scaling Still ImprovesDensity, Performance, Power, Cost

130 nm 90 nm130 nm 90 nm MadisonMadison MontecitoMontecitoCores/ThreadsCores/Threads 1/11/1 2/4 2/4TransistorsTransistors 0.410.41 1.721.72 BillionBillionL3 CacheL3 Cache 66 2424 MByteMByteFrequencyFrequency 1.51.5 >1.7>1.7 GHzGHzRelative PerformanceRelative Performance 11 >1.5x>1.5xThermal Design Power Thermal Design Power 130130 ~100~100 WattWatt


Gate Count

Design Cost

New tech. node model

R&D Cost

Facilities development cost

Revenues

Revenue

Key Points

• Moore’s Law thriving after 40 years

• Convergence drives IC industry growth

• Integrated platforms optimize user experience

• Multi-core parallelism going mainstream

Convergence Drives Growth

ConvergenceConvergence

1010

00

5050

100100

20032003

6565

90909696

20042004 20052005 20062006

% of Notebooks % of Notebooks that have Wireless that have Wireless

ConvergenceWireless Mobile PC

Source: IDCSource: IDC

Source: Goldman Sachs and Co., McKinsey & Company; Dec. 2002

Bits

of T

raffi

cB

its o

f Tra

ffic

ConvergenceData Phones and Traffic

2004: Data Phones 2004: Data Phones Cross-over VoiceCross-over Voice

Source: WebFeet Research

New Cell Phone Sales by Feature SetNew Cell Phone Sales by Feature Set

00

100100

200200

300300

400400

500500

600600

700700

800800

Mill

ion

units

Mill

ion

units

2009200920022002 20032003 20042004 20052005 20062006 20072007 20082008

DataDataVoiceVoice

Data Traffic: 56% Annual Data Traffic: 56% Annual Growth thru 2006Growth thru 2006

DataData

VoiceVoice

20012001 20062006

Sources: Consumer Electronics Association, Photo Marketing Association International, Photofinishing News, SBCA/Sky Trends

’’9595 ’96’96 ’97’97 ’98’98 ’99’99 ’00’00 ’01’01 ’02’02 ’03’03 ’04’04 ’05’05 ’06’06 ’07’07 ’08’08

The Great CrossoverThe Great CrossoverDigital Technologies Surpass AnalogDigital Technologies Surpass Analog

Crossover YearCrossover Year

ConvergenceDigital Consumer Electronics

Camera SalesCamera SalesMillionsMillions

‘‘9595 ‘‘9696 ‘‘9898‘‘9797 ‘‘9999 ‘‘0000 ‘‘0101 ‘‘0202 ‘‘0303 ‘‘0404 ‘‘0505Portable CD PlayersPortable CD Players

DIGITAL

ANALOG

SatellitesSatellites

Cell PhonesCell Phones

DVD PlayersDVD PlayersCamcordersCamcorders

CamerasCamerasTVsTVs

Key Points





Demand Growth Driven by Better User Experiences

MemoryMemoryComputingComputing

GraphicsGraphicsNetworkingNetworking

WirelessWirelessMultimediaMultimedia

MobilityMobilityCompactnessCompactness

Battery LifeBattery Life

VirtualizationVirtualizationSecuritySecurity

MultitaskingMultitasking

Ease of UseEase of UseManageabilityManageability

UserUser ExperienceExperience

Platforms Optimize User Experience

Innovate and IntegrateInnovate and Integrate

MathMathCoprocessorCoprocessor SSE2SSE2

ItaniumItaniumArchitectureArchitecture

Low Low powerpower

Multi CoreMulti CoreHyper-Hyper-

ThreadingThreading

ProcessorProcessorCapabilitiesCapabilities

IntegratedIntegratedCacheCache PCIPCI

AGP 2XAGP 2X

USBUSB

AGP 4XAGP 4XIAPCIAPC

USB 2.0USB 2.0AGP 8XAGP 8X

Serial ATASerial ATA HD AudioHD AudioInfinibandInfiniband

PCI-EPCI-EGraphicsGraphics

Low PowerLow PowerChipsetChipsetCapabilitiesCapabilities

UserUser ExperienceExperience

MultipleMultipleExecution UnitsExecution Units

MMXMMXSSESSE

QuickstartQuickstart

SpeedstepSpeedstep

BaseBasePerformancePerformance

VirtualizationVirtualization

ProSetProSet

WiFiWiFi Comms andComms andSoftwareSoftwareCapabilitiesCapabilities

386386processorprocessor

Pentium® Pentium® processorprocessor

Pentium®II Pentium®II processorprocessor

Pentium®III Pentium®III processorprocessor

Pentium®IIIPentium®IIIXeon™ Xeon™

processorprocessor

Itanium™ Itanium™ processorprocessor

Pentium®III Pentium®III mobilemobile

processorprocessorPentium®4 Pentium®4 processorprocessor

Pentium®4 Pentium®4 Processor Processor

with HTwith HT

Pentium®4 Pentium®4 mobilemobile

processorprocessor

Xeon™Xeon™processorprocessor

ItaniumItanium™™2 2 processorprocessor

Example: Intel platformsExample: Intel platforms

RMS Applications Growing

Emerging workloads increase need forEmerging workloads increase need forhigh performance parallel processinghigh performance parallel processing

Performance and Power Efficiency

Increase with Parallel ArchitectureRelative Relative

processor processor performance*performance*(constant power (constant power

envelopeenvelope))


*Average of SPECInt2000 and SPECFP2000 rates for Intel desktop processors *Average of SPECInt2000 and SPECFP2000 rates for Intel desktop processors vs initial Intelvs initial Intel®® Pentium Pentium®® 4 Processor 4 Processor

FORECASTFORECAST

1

10

100

20002000

Single-CoreSingle-Core

2008+2008+20042004

10X10X

Dual / Multi-CoreDual / Multi-Core

3X3X

Multi-Core Parallelism Going Mainstream

Dual-Core Processor Plans (Intel)Dual-Core Processor Plans (Intel)ServersServers DesktopDesktop MobileMobile

90nm90nm MontecitoMontecito

(2005)(2005)

65nm 65nm Desktop ProcessorDesktop Processor

(2006)(2006)

65nm 65nm YonahYonah(2005)(2005)

90nm 90nm Server ProcessorServer Processor

(2006)(2006)

90nm 90nm SmithfieldSmithfield

(2005)(2005)

Note: Die images not to scale and may not depict actual productNote: Die images not to scale and may not depict actual product


Key Points





• Holistic solutions deliver power efficiency

Silicon Technology Changes to Increase Power Efficiency

• 1960’s: Bipolar

• 1970’s: PMOS, NMOS

• 1980’s: CMOS

• 1990’s: Voltage scaling (P = CV2f)

• 2000’s: Power efficient scaling/design

High High Stress Stress

FilmFilm

NMOSNMOS

SiGeSiGe SiGeSiGe

PMOSPMOS

Compressive channel strain Tensile channel strainCompressive channel strain Tensile channel strain30% drive current increase 10% drive current increase30% drive current increase 10% drive current increase

Power Efficient 90nm Transistors with Strained Silicon

Innovate and integrateInnovate and integratefor cost effective production for cost effective production

Source: Intel

does not raise transistor leakage

11

1010

100100

10001000

0.20.2 0.40.4 0.60.6 0.80.8 1.01.0 1.21.2 1.41.4 1.61.6Transistor Drive Current (mA/um) Transistor Drive Current (mA/um)

StdStd StdStdStrainStrain StrainStrain

+25% I+25% I ONON +10% I+10% I ONON

0.04x I0.04x I OFFOFF

0.20x I0.20x I OFFOFF

PMOSPMOS NMOSNMOS

Strained Silicon Improves Transistor Performance and/or

Reduces Leakage

TransistorTransistorLeakageLeakageCurrentCurrent(nA/um)(nA/um)

Source: Intel

Using Using priorprior

designdesigntechniquestechniques

With new With new powerpower

reductionreductiontechniquestechniques

ISSCC 2005 P10.1ISSCC 2005 P10.1““The Implementation of The Implementation of a 2-core Multi-Threadeda 2-core Multi-Threaded

ItaniumItaniumTMTM Family Family Processor”Processor”

Power Power (W)(W)

Advances in Power Efficient Design

Circuit Techniques Reduce Source Drain Leakage

Body BiasBody Bias

+ Ve+ Ve

VddVdd VbpVbp

VbnVbn

- Ve- Ve

2 - 10X2 - 10X

Sleep TransistorSleep Transistor

2 - 1000X2 - 1000X

Stack EffectStack Effect

2 - 10X2 - 10X

Logic Logic BlockBlock

Equal LoadingEqual Loading

Leakage Leakage ReductionReduction

Sleep Transistor Reduces SRAM Leakage Power

VVSSSS

VVDDDD

NMOS NMOS Sleep Sleep

TransistorTransistor

SRAM SRAM Cache Cache BlockBlock

70 Mbit SRAM leakage current map70 Mbit SRAM leakage current map

Without sleep transistorWithout sleep transistor With sleep transistorWith sleep transistor

Accessed blockAccessed block

>3x SRAM leakage reduction on inactive blocks>3x SRAM leakage reduction on inactive blocks

Source: Intel

Sleep Transistors Reduce ALU Leakage

Scan Scan FIFOFIFO

Scan Scan outout

Sleep Sleep ALUALU

Body biasBody bias

ControlControl

37X leakage reduction 37X leakage reduction demonstrated on test chipdemonstrated on test chip

Scan Scan

PMOS underdrive VPMOS underdrive Vcccc

PMOS overdrive VPMOS overdrive Vssss

Virtual VVirtual Vcccc

Virtual VVirtual Vssss

ScanScancapturecapturecontrolcontrol

DynamicDynamicALUALU 3232

NMOS overdrive VNMOS overdrive Vcccc

NMOS underdrive VNMOS underdrive Vssss

ALU coreALU core

3-bit A/D3-bit A/D

ALUALUBodyBodyBiasBias

PMOSPMOSSleepSleepBodyBodyBiasBias

VVssss externalexternal

VVcccc

externalexternal

Sleep transistorSleep transistorand body biasand body bias

controlcontrol

Sleep transistorsSleep transistors

Source: ISSCC 2003, Paper 6.1

What happens to Analog guys?

In the nanoelectronics era CMOS performance will exceed that of the SiGe HBT

Holistic Approach to System Power Projected Change in Peak Power Consumption Projected Change in Peak Power Consumption

of High-End Desktop Computersof High-End Desktop Computers

00

0.50.5

11

1.51.5

22

2.52.5

33

20022002 20042004 2005 (est.)2005 (est.)

Po

wer

Co

nsu

mp

tio

n R

elat

ive

to a

200

2 P

C

Po

wer

Co

nsu

mp

tio

n R

elat

ive

to a

200

2 P

C

Power SupplyPower Supply

OtherOther

VRMsVRMs

Graphics CardGraphics Card

ProcessorProcessor

Projected power supply efficiencyProjected power supply efficiency

20%20%

30%30%

43%43%

10%10%

23%23%

41%41%

57%57% 70%70% 80%80%

Source: Intel

Key Points





• Holistic solutions deliver power efficiency

• Nanotechnology will extend IC advances

• Lithography innovations remain vital

Silicon Technology Reaches Nanoscale

MicronMicron

1000010000

10001000

100100

1010

1010

11

0.10.1

0.010.01

Nano-Nano-metermeter

NanotechnologyNanotechnology(< 100nm)(< 100nm)

130nm130nm

90nm90nm

70nm70nm 50nm50nm

Gate LengthGate Length 65nm65nm

35nm35nm

19701970 19801980 19901990 20002000 20102010 20202020

45nm45nm32nm32nm

22nm22nm

25nm25nm 18nm18nm

12nm12nm

0.7X every 2 years

Nominal feature sizeNominal feature size

Source: Intel

• Structures measured in nanometers– Less than 0.1-micron (100nm)

• New processes, materials, device structures– Incrementally changing silicon technology base

• Materials manipulated on atomic scale– In one or more dimensions

• Increasing use of self-assembly– Using chemical properties to form structures

Nanotechnology innovations will extendNanotechnology innovations will extendsilicon technology and Moore’s Lawsilicon technology and Moore’s Law

Nanotechnology Hallmarks(For Nanoelectronics)

Step 3Step 3

Step 4Step 4 Step 2Step 2

AA

Step 1Step 1

AAAA

BB

AABB

Design Your Own Film withAtomic Layer Deposition (ALD)

Atomic level manipulation + Self-assemblyAtomic level manipulation + Self-assembly

ALD Enables High-k Dielectric to Reduce Gate Leakage

Silicon substrate

Gate

3.0nm High-k

Silicon substrate

1.2nm SiO2

Gate

High-k vs. SiO2 Benefit

Gate capacitance 60% greater Faster transistors

Gate dielectric leakage

> 100x reduction Lower power

Source: Intel

Process integration is the key challengeProcess integration is the key challenge

Nanostructures for the Next Decade

(Transistor Research at Intel)

Si DeviceSi DeviceMiniaturizationMiniaturization

Non-planarNon-planarTri-GateTri-GateArchitectureArchitecture

III-V DeviceIII-V DeviceResearchResearch

Carbon Carbon NanotubeNanotubeTransistorTransistor

Gate

Source

Sibody

(b)

Drain

Source

CNTSingle-wallD = 1.4 nm

Gate(Pt)

Drain(Pd)

Lg = 75 nm

(d)

Gate

DrainSource

Source

(c)

Multiepitaxiallayers

Gate

Source

Sibody

(b)

Drain

Gate

Source

Sibody

(b)

Drain

Source


Gate(Pt)

Drain(Pd)

Lg = 75 nm

(d)

Source


Gate(Pt)

Drain(Pd)

Lg = 75 nm

(d)

Gate

DrainSource

Source

MultiepitaxiallayersGate

DrainSource

Source

Multiepitaxiallayers

(c)

(a)

L = 10 nmg

GateGateSourceSource

DrainDrain

Source: Intel

0.1

1

10

100

1 10 100 1000 10000GATE LENGTH LG [nm]

GATE D

ELA

Y CV/I

[p

s]

Si MOSFETs

CNT FETs

III-V FETs

NMOS

Benchmarking Nanotransistor Progress

Source: Intel compilation of published dataSource: Intel compilation of published data

Lithography Must Break Throughto Shorter Wavelength (EUV* @ 13.5nm)

* Extreme Ultraviolet* Extreme Ultraviolet

Source: Intel

10001000

100100

1010’’8989 ’’9191 ’’9393 ’’9595 ’’9797 ’’9999 ’’0101 ’’0303 ’’0505 ’’0707 ’’0909 ’’1111

Feature sizeFeature size

EUVEUV

Lithography Lithography

WavelengthWavelength

193nm & extensions193nm & extensions248nm248nm

nmnm

GapGap

EUV Lithography inCommercial Development

EUV Micro exposure tool (MET)EUV Micro exposure tool (MET)EUV MET Image (8/04)EUV MET Image (8/04)

• Integrated development in progress

• Source power and lifetime

• Defect free mask fabrication and handling

• Optics lifetime

• Resist performance

Source: Intel

Key Points

• Moore’s Law thriving after 40 years• Convergence drives IC industry growth• Integrated platforms optimize user experience• Multi-core parallelism going mainstream• Holistic solutions deliver power efficiency• Nanotechnology will extend IC advances• Lithography innovations remain vital• Moore’s Law will outlive CMOS• Future rides on innovation and integration

BipolarBipolar

Moore’s Law Will Outlive CMOS

1960 1970 1980 1990 2000 2010 2020 2030

CMOSCMOS

1µm 100nm 10nm10µm

PMOSPMOS

101088

101077

101066

101055

101044

101033

101022

101011

101000

101099

10101010 1965 Data (Moore)1965 Data (Moore)

MicroprocessorMicroprocessorMemoryMemory

10101111

10101212

10101313 VoltageVoltageScalingScaling

Pwr EffPwr EffScalingScaling

New Nano-New Nano-structuresstructures

NMOSNMOS

KiloXtor

MegaXtor

GigaXtor

TeraXtor

Tra

nsi

sto

rs/D

ieT

ran

sist

ors

/Die

Beyond CMOS?Beyond CMOS?Spin based?Spin based?Molecular?Molecular?Other?Other?

Innovation and IntegrationWill Sustain Moore’s Law

Identify needs andIdentify needs andcreate capabilities create capabilities that drive growththat drive growth

Deliver platforms Deliver platforms to optimize user to optimize user experienceexperience

Anticipate barriers Anticipate barriers and seek timely and seek timely breakthroughsbreakthroughs

Make strategic Make strategic technology technology transitionstransitions

Integrate new Integrate new materials, devices, materials, devices, processesprocesses

Coordinate Coordinate strategic shifts strategic shifts across industryacross industry

InnovationInnovation IntegrationIntegration

Introduction Mohammad Sharifkhani. The IC Growth CycleDensityPerformance Switching energy Cost per...

Documents

Transcript of Introduction Mohammad Sharifkhani. The IC Growth CycleDensityPerformance Switching energy Cost per...