Copyright Gordon Bell & Jim Gray Computing Laws Beyond Moores Law The best way to predict the future is to invent it. --Alan Kay Gordon Bell Bay Area Research.

Copyright Gordon Bell & Jim Gray Computing LawsCopyright Gordon Bell & Jim Gray Computing Laws

Beyond Moore’s Law

The best way to predict the future is to invent it. --Alan Kay

Gordon Bell

Bay Area Research Center

Microsoft Corporation


Beyond Moore’s Law

Just FCB (faster, cheaper, better)… COTS will soon mean consumer off the shelf

Moore’s Law and technology progress likely to continue for another decade for: processing, memory, storage, LANs, WANs

System-on-a chip of interesting sizes will emerge to create 0 cost systems

Any displacement technology is unlikely… Carver Mead’s Law c1980A technology takes 11 years to get established

On the other hand, we are on Internet time! No DNA, molecular, or quantum computers, or new stores


Beyond Moore’s Law Results

Is the Internet aka www.everything? Moore’s Law to get cheaper, one chip

systems that increase portability, ubiquity, etc.

Paper-competitive Screens Disks of 1 TB Wireless for ubiquity; including GPS Bridges to television Bridges to PSTN for phones, PDAs, etc.


Beyond Moore’s Law Results The more uniform the system,

the more attractive it is for developers to produce many varieties of low cost apps

The more uniform the system, the more susceptible they are to viruses

Change will be due to ubiquity of computing brought about by networking

PLUS Interesting, new platforms that interface use/users

– When can we speak to these computers?– Sensors e.g. cameras of all types– GPS and direction (pointing)– MEMS & Biochips in particular

There are many other laws and forces, beyond Moore’s Law that determine IT


Big event of 1999: massive infusion of venture capital

>$3 Billion/quarter (1/3 for Internet).…Esprit $3B/3 yrs

Capital is pulling people from research. Product development beats research if you

have an idea what you’re looking for Little technology. Apps development. 1960-2000: shift from central to distributed

back to fully distributed computing


Forecast of corp web-enabled expenditures

In a decade we can/will have: more powerful personal computers

– processing 10-100x– 4x resolution (2K x 2K) displays to impact paper– Large, wall-sized and watch-sized displays– low cost, storage of one terabyte for personal use

adequate networking????– ubiquitous access = today’s fast LANs– Competitive wireless networking

One chip, networked platforms including light bulbs, cameras everywhere, etc.

Some well-defined platforms that compete with the PC for mind (time) and market sharewatch, pocket, body implant, home

Inevitable, continued cyberization… the challenge… interfacing platforms and people.


What if could or when can we store everything we’ve: read/written, heard, and seen?



Storing all we’ve read (written), heard (listened to), & seen (presented)Human data-types /hr /day (/4yr) /lifetimeread text, few pictures 200 K 2 -10 M/G 60-300 G

speech text @120wpm 43 K 0.5 M/G 15 Gspeech @1KBps 3.6 M 40 M/G 1.2 T

video-like 50Kb/s POTS 22 M .25 G/T 25 Tvideo 200Kb/s VHS-lite 90 M 1 G/T 100 T

video 4.3Mb/s HDTV/DVD 1.8 G 20 G/T 1 P


High Performance Computing

Copyright Gordon Bell & Jim Gray Computing LawsCopyright Gordon Bell & Jim Gray Computing Laws0.0001

0.001

0.01

0.1

1

10

100

1000

1985 1990 1995 2000 2005 2010

Bell Prize and Future Peak Tflops (t)

Petaflops study target

NEC

XMP NCube

CM2

*IBM

Computer types

NetwrkedSupers…

GRIDLegionCondor Beowulf NT clusters

VPPuni

T3E SP2(mP) NOW

NEC mP

SGI DSM clusters &SGI DSM

NEC super Cray X…T(all mPv)

MainframesMultis

WSs PCs

-------- Connectivity--------

WAN/LAN SAN DSM SM

mic

ros

v

ecto

r

Clusters


High Performance Computing

Supers we knew are Japanese; scalability & COTS in… but you have to roll your

own else pay the Unix & proprietary taxes Beowulf is $14K/TB ( 6 x 4 x 40 GB) IBM 4000R 1 rack: 2x42 500Mhz processors, 84

GB, 84 disks (3TB @36GB/disk)$420K … still cheaper than the “big buys”

$10-20K/node for special purpose vs $2K for a MAC

EMC, IBM at $1 million/TB; vs $14K

Region/Region/IntranetIntranet

CampusCampusHome…Home… buildingsbuildings

BodyBody

WorldWorld

ContinentContinent

Everything cyberizable will be Everything cyberizable will be in Cyberspace and covered in Cyberspace and covered by a hierarchy of computers!by a hierarchy of computers!

Fractal Cyberspace: a network of … networks of … platforms

Cars… Cars… phys. nets phys. nets


Cyberization: interface to all bits and process information

Coupling to all information and information processors

Pure bits e.g. printed matter Bit tokens e.g. money State: places, things, and people State: physical networks


Bell’s law of computer class formation to cover Cyberspace

New computer platforms emerge based on chip density evolution

Computer classes require new platforms, networks, and cyberization

New apps and content develop around each new class

Each class becomes a vertically disintegrated industry based on hardware and software standards

Bell’s Evolution Of Computer Classes

Technology enables two evolutionary paths:1. constant performance, decreasing cost2. constant price, increasing performance

1.26 = 2x/3 yrs -- 10x/decade; 1/1.26 = .81.26 = 2x/3 yrs -- 10x/decade; 1/1.26 = .81.6 = 4x/3 yrs --100x/decade; 1/1.6 = .621.6 = 4x/3 yrs --100x/decade; 1/1.6 = .62

MiniMini

TimeTime

Mainframes (central)Mainframes (central)

PCs (personals)PCs (personals)Lo

g p

rice

Lo

g p

rice

WSsWSs

HandheldHandheld ?? ??


Platform evolution: What do they do that’s useful?

How do they communicate?

Price, performance, and class of various goods & services

Computer price = $10 x 10 class#

Computer weight = .05 x 10 class#

Car price = $6K x 1.5 class #

Transportation artifact prices = k x $10 type (shoes,...cars,... trains,... ICBMs)

French Restaurants(t='95) =f(ambiance, location) x $25 x 1.5 stars

•

Bell’s Ten+ Computer Price Tiers

Super server: costs more than $100,000“Mainframe”: costs more than $1 million

an array of processors, disks, tapes, comm ports

1$: embeddables e.g. greeting card 10$: wrist watch & wallet computers

100$: pocket/ palm computers 1,000$: portable computers 10,000$: personal computers (desktop) 100,000$: departmental computers

(closet) 1,000,000$: site computers (glass house) 10,000,000$: regional computers (glass

castle) 100,000,000$: national centers 1,000,000,000$: the grid


On body and in body networks

Third wearables conference


Libretto, .5mm

PCS; Pilot

Libretto PS, Ricoh Camera; Swiss Army Knife

Compass; altimeter

Not shown: ECG; GPS;


22 years ago: 6 oz. Watch, manual size > watch size


Audio, pix, T, P, ECG, location, physiological parameters…1 GB


Steve ManninCyberspace


CMU wearable computers


MedronIc


Your husband just died, … here’s his black box

When will we have smart rooms? Reasonable sized displays or panel for

interaction Cameras that can recognize various people Mics and Speech based interface Speakers Coupled to all power, data, audio, and

video/television networks

Interval Research has a product to track individuals in stores!


Or be completely covered by a smart world

450 Old Oak Ct, Los Altos, CA


Webcams


Webcam of Hospital in Sweden


Economics-based laws determine the market As industries increase, they become horizontal Demand: doubles as price declines by 20% Learning curves: 10-15% cost decline with 2X units Nathan’s Laws of Software -- the virtuous circle Bill’s Law for the economics of PC software Linus’s Law for software… it is free plus support Sarnoff & Metcalf Laws for the “value of a network”


Consult

Apps

Apps

Dbases

OS

Network

Periph

Computers

Micros

Solutions

Andersen, EDS, KPMG, Lante, etc.Andersen, EDS, KPMG, Lante, etc.

Microsoft, Lotus, WordPerfect, etc.Microsoft, Lotus, WordPerfect, etc.

Microsoft, Apple, Sun, NovellMicrosoft, Apple, Sun, Novell

Comshare, D&B, PeopleSoft, SAPComshare, D&B, PeopleSoft, SAP

HP, Canon, Lexmark, SeagateHP, Canon, Lexmark, Seagate

Novell, Microsoft, BanyanNovell, Microsoft, Banyan

IBM, Compaq, DEC, Apple, many othersIBM, Compaq, DEC, Apple, many others

Intel, AMD, Motorola, othersIntel, AMD, Motorola, others

Informix, Ingres, Oracle, Sybase,etc.Informix, Ingres, Oracle, Sybase,etc.

EDS, FDC, BTG, API, DataFocus, HFSIEDS, FDC, BTG, API, DataFocus, HFSI

Computer Industry 1995

Applications

Applications

Databases

OS

Switching

Computers

DSP

Processors

Microsoft, Delrina, many othersMicrosoft, Delrina, many others

Microsoft, Apple, Sun, Novell, LINUXMicrosoft, Apple, Sun, Novell, LINUX

Ericsson, Aspect, Nortel, Octel, othersEricsson, Aspect, Nortel, Octel, others

Dialogic, NMS, Rhetorex, othersDialogic, NMS, Rhetorex, others

Ericsson, Nortel, Bay, 3Com, Fore, othersEricsson, Nortel, Bay, 3Com, Fore, others

Compaq, DEC, Dell, IBM, many othersCompaq, DEC, Dell, IBM, many others

Intel, AMD, Motorola, othersIntel, AMD, Motorola, others

Informix, Microsoft, Oracle, Sybase, othersInformix, Microsoft, Oracle, Sybase, others

Future Telecom Industry


Internet Industry (circa 1999)Courtesy of Zindigo Ventures

TransportTransportNetwork Hardware/ProtocolsNetwork Hardware/Protocols

Computers & Operating Layer SoftwareComputers & Operating Layer SoftwareApplications & MiddlewareApplications & Middleware

Info

ba

se s

/Po

r ta l

sIn

f ob

as

e s/P

ort

als

Pro

cur e

men

tP

r ocu

r em

ent

Co

mm

un

i ca t

ion

Co

mm

un

i cat

ion

Su

pp

l y C

hai

nS

up

ply

Ch

a in

ER

PE

RP

Pro

f ess

i on

a lP

r of e

ssi o

nal

Fin

anci

a lF

inan

cial

Op

era t

ion

sO

per

atio

ns

Mar

ket i

ng

Mar

ket i

ng

Internet Services$170B*

Infrastructure$171B*

AccessAccessPersonal/Employee DataPersonal/Employee Data

Go

vern

men

tG

ove

rnm

ent

Content SyndicatorsContent SyndicatorsContent Syndication

$2B+ **

Web HostingWeb Hosting


Nathan’s Laws of software

1. Software is a gas. It expands to fill the container it is in

2. Software grows until it becomes limited by Moore’s Law

3. Software growth makes Moore’s Law possible

4. Software is only limited by human ambition and expectation

…GB: and our ability to cyberize I.e. encode

Software Economics: Bill’s Law

Bill Joy’s law (Sun): don’t write software for <100,000 platforms @$10 million engineering expense, $1,000 price

Bill Gate’s law:don’t write software for <1,000,000 platforms @$10M engineering expense, $100 price

Examples: –UNIX versus Windows NT: $3,500 versus $500–Oracle versus SQL-Server: $100,000 versus $6,000–No spreadsheet or presentation pack on UNIX/VMS/...

Commoditization of base software and hardware

PricePriceFixed_costFixed_cost

Marginal _costMarginal _cost==UnitsUnits

++

Copyright Gordon Bell & Jim Gray Computing LawsCopyright Gordon Bell & Jim Gray Computing LawsInno

vatio

n

The Virtuous Economic Cycle that drives the PC industry

Volum

e

Competition

Standards

Utility/value

Linus’s Law: Linux everywhere

Software is or should be free All source code is “open” Everyone is a tester Everything proceeds a lot faster when

everyone works on one code Anyone can support and market the code for

any price Zero cost software attracts users! All the developers write lots of code


Sarnoff’s Law

The value of a network is proportional to the number of its users


Metcalf’s LawNetwork Utility = Users2

How many connections can it make?– 1 user: no utility– 100,000 users: a few contacts– 1 million users: many on Net– 1 billion users: everyone on Net

That is why the Internet is so “hot”– Exponential benefit

Telnet & FTP

EMAIL

WWW Audio Video

Voice!Voice!

StandardsStandards

Increase Capacity(circuits & bw)

Lower response time

Create newservice

Increased Demand

The virtuous cycle of bandwidth supply and demand


What is the value of combined network when television, telephone, and hand

held web devices are added?

How do you build a home network infrastructure,

platforms, and interface to uses

HomeCATV

Analog/digital cable distribution

PC broadcasts are mixed into home CATV in analog and/or MPEG digital

Ethernet Home network

Video capture

“milliBill”

Basic ideas:

1. PC records or plays thru video cable channels. 2. PC “broadcasts” art images, webcams, presentations,

videos, DVDs, etc.3. Ethernet not cable?

Settopbox

Another big bang? Internet to TV and audio: The Net, PC meet the TV


PCTV a.k.a. MilliBillgUsing PCs to drive large screens e.g. tv sets, Plasma Panels

Gordon BellJim Gemmell

Bay Area Research CenterMicrosoft Research

Copyright 1999 Microsoft Corporation




Voice to WEBVoice to WEBBridgeBridge

Web ServerWeb Server

TheTheWebWeb

DataBaseDataBase

PSTNPSTN

The Next ConvergencePOTS connects to the Weba.k.a. Phone-Web Gateways


Enable voice and text access on phones, screen phones, PDAs and other devices to existing Internet infrastructure in an intelligent, customizable way.

WebOnPhonWebOnPhonee

Mission:Mission:

WebOnPhone


DataData

Cyberspace: one, two or three networks? in 2005, 2010, 2020

TelephonyTelephony

TelevisionTelevision

Will we have gateways?


Hardware technology: processing, memory, networking, and new interfaces enable the new computers


1. We get more


Tera

Giga

Mega

Kilo

11947 1957 1967 1977 1987 1997 2007

Extrapolation from 1950s: 20-30% growth per year

StorageStorageBackboneBackbone

MemoryMemoryProcessingProcessing

Telephone ServiceTelephone Service17% / year17% / year

????


National Semiconductor Technology Roadmap (size)

1

10

100

1000

10000

1995 1998 2001 2004 2007 2010

Mem

ory

siz

e (

Mb

yte

s/c

hip

) &

Mtr

an

sis

tors

/ ch

ip

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

Mem(MBytes)

Micros Mtr/chip

Line width


National Storage Technology Roadmap (size, density, speed)

1

10

100

1000

10000

100000

1995 2000 2005

1

10

100

1000

10000

1000003.5" Cap. (By tes )

1.3" Cap. (By tes )

Bits /s q. in.

Data-rate (By tes /s )


0.01

0.1

1

10

100

1000

10000

1986

1988

1990

1992

1994

1996P

erf

orm

an

ce in

Mfl

op

/s

Micros

Supers

8087 802876881

80387

R2000

i860

RS6000/540Alpha

RS6000/590Alpha

Cray 1S

Cray X-MP

Cray 2 Cray Y-MP Cray C90Cray T90

1998

Growth of microprocessor performance

1980

1982


Microprocessor performance

100 G

10 G

Giga

100 M

10 M

Mega

Kilo1970 1980 1990 2000 2010

Peak Peak Advertised Advertised

Performance Performance (PAP)(PAP)

Moore’sMoore’sLawLaw

Real AppliedReal AppliedPerformance Performance

(RAP) (RAP) 41% Growth41% Growth

System-on-a-chip alternatives

FPGA Sea of un-committed gate arrays

Xylinx, Altera

Compile a system

Unique processor for every app

Tensillica

Systolic | array

Many pipelined or parallel processors

DSP | VLIW

Special purpose processors

TI

Pc & Mp.

ASICS

Gen. Purpose cores. Specialized by I/O, etc.

Intel, Lucent, IBM

Universal Micro

Multiprocessor array, programmable I/o

Cradle

Cradle: Universal Microsystemtrading Verilog & hardware for C/C++

Single part for all apps Programming @ run time via FPGA & ROM 5 quad mPs at 3 Gflops/quad = 15 Glops Single shared memory space, caches Programmable periphery including:

1 GB/s; 2.5 GipsPCI, 100 baseT, firewire

$4 per flops; 150 mW/Gflops

UMS : VLSI = microprocessor : special systemsSoftware : Hardware

MSP

MSP

MSP

M EM O R Y

MSP

MSP

MSP

MSP

M EM O R Y

MSP

MSP

MSP

MSP

M EM O R Y

C LO C KS,D EBU G

MSP

MSP

MSP

MSP

M EM O R YD R AMC O N TR O L

MSP

D R AM

PR O G I/O PR O G I/O PR

OG

I/O

PR

OG

I/O

PR

OG

I/O

PROG I/OPROG I/OPROG I/OPROG I/O

PR

OG

I/OP

RO

G I/O

PR

OG

I/O

N VM EM

UMS Architecture

Memory bandwidth scales with processing Scalable processing, software, I/O Each app runs on its own pool of processors Enables durable, portable intellectual property


Gains if 20, 40, & 60% / year

1.E+21

1.E+18

1.E+15

1.E+12

1.E +9

1.E+61995 2005 2015 2025 2035 2045

20%= 20%= TeraopsTeraops

40%= 40%= PetaopsPetaops

60%= 60%= ExaopsExaops


Communication rate(t) in log10(Kbps)

200519951985197519652

3

4

5

6

7

8

9

10

POTS

WAN

LAN

SAN/backpanels

1 Mb

1 Gb

1 Kb

???

???

POTS @17%/year

ISDN


Nomadicity

Video... Plus >>B/W

Universality

USA Today 1 Sept. 99

Copyright Gordon Bell & Jim Gray Computing LawsCopyright Gordon Bell & Jim Gray Computing Laws0

0.2

0.4

0.8

2.0

199719981999200020012002

Circuit data<9.6kbps

HSCSD57.6kbps

GPRS115kbps

EDGE384kbps

UMTS2Mbps

0.1

The evolution of wireless data standards

T1, T3, …T1, T3, …

Public SpacesDiscovery of proximity services (flight schedules, mall directories)

Proxy Proxy ServerServer

PhonePhone

EthernetEthernet

Internet

WebWebServerServer 802.11802.11

IrDAIrDACellularCellular

BluetoothBluetooth

Copyright Gordon Bell & Jim Gray Computing LawsCopyright Gordon Bell & Jim Gray Computing Laws1970 1975 1980 1985 1990 1995 2000 2005 2010

10 Tbps

100 Tbps

1 Tbps

100 Gbps

10 Gbps

1 Gbps

100 Mbps

10 Mbps

1 Mbps

100 Kbps

10 Kbps

1 Kbps

100 bps

10 bps

1 Pbps

Voice Traffic

56 KB

T1T3

OC-3OC-12

OC-48OC-192

OC-768

Voice Crossover

Internet TrafficMax. Port Speed

$100 M

$10 M

$1 M

$100 K

$100 K

1997 Breakpoint

Delay Engineered

Capacity E

ngineere

d

Max. Switch Speed

100 Pbps

10 Pbps

Internet growths vs timecourtesy of Dr. Larry Roberts


Desktop-desktop @ 1 gbps

2x 1GbE3x 1GbE

G8 Sonet

SC99NOCGigaPop

Juniper M40Cisco 12008

WestinGigaPop

Juniper M40

G2 SonetOC12 Sonet

Cisco 1800

G2 SonetOC12 Sonet

Juniper M40 NCSAUW

University of Washington Microsoft

Nortel NetworksNTON

SC99 @ Portland

1 Gbps5x 270 Mbps

HDTV streams= 1 Gbps

OC-48 = 2,488.32 Mbps = G2OC192 = 9,953.28 Mbps = G8

2x 1GbE3x 1GbE

G8 Sonet

SC99NOCGigaPop

Juniper M40Cisco 12008

WestinGigaPop

Juniper M40

G2 SonetOC12 Sonet

Cisco 1800

G2 SonetOC12 Sonet

Juniper M40 NCSAUW

University of Washington Microsoft

Nortel NetworksNTON

SC99 @ Portland

1 Gbps1 Gbps5x 270 Mbps

HDTV streams= 1 Gbps

OC-48 = 2,488.32 Mbps = G2OC192 = 9,953.28 Mbps = G8

http://research.microsoft.com/~gray/papers/Win2K_1Gbps.doc

1988 Federal Plan for Internet

In a decade we can/will have: more powerful personal computers

– processing 10-100x– 4x resolution (2K x 2K) displays to impact paper– Large, wall-sized and watch-sized displays– low cost, storage of one terabyte for personal use

adequate networking????– ubiquitous access = today’s fast LANs– Competitive wireless networking

One chip, networked platforms including light bulbs, cameras everywhere, etc.

Some well-defined platforms that compete with the PC for mind (time) and market sharewatch, pocket, body implant, home

Inevitable, continued cyberization… the challenge… interfacing platforms and people.


The End


Things get cheaper


Exponential change of 10X per decade causes real turmoil!100000

10000

1000

100

$K 10

1

0.1

0.01 1960 1970 1980 1990 2000

8 MB8 MB

1 MB1 MB

256 KB256 KB

64 KB64 KB

16 KB16 KB

Timeshared Timeshared systemssystems

Single-userSingle-usersystemssystems


VAX Planning Model 1975:I didn’t believe it The model was very good

– 1978 timeshared $250K VAXencost about $8K in 1997!

Costs declined > 20% – users get more memory than predicted

Single user systems didn’t come down as fast, unless you consider PDAs

VAX ran out of address bits!


Newer & cheaper always wins?… if it weren’t for the Law of Intertia

Old

NewNew

Old

NewNew


“The mainframe is dead!… and for sure this time!”

PRICE

MainframeMainframe

ServerServer

PCPC


The law of data and program inertia sustains platforms!

The investment in programs and processes to use them, and data exceed hardware costs

The cost to switch among platforms e.g. IBM mainframe, VMS, a VendorIX, or Windows/NT is determined by the data and programs

The goal of hardware suppliers is uniqueness to differentiate and lock-in

The goals of software/database suppliers are: to differentiate and lock-in and operate on as many platforms as possible in order to be not tied to a hardware vendor


Computer industry growth(Gbell’s swag 12/99)

Machine class 1992 1998 2004Watch >Cellphone WAP >>Appliance of some type =TC (TV Computer) na = >>Handhelds >> =Network Computer - = >PC (portables) > > >PC (desktop) = = =Workstation = < <VendorIX server (mini) > >> >Mainframe < < <Super (classic) = < <<Scalable PCs = > >>= 0-10%, >10-20%, >> 20-30%; < -10%


The End

Copyright Gordon Bell & Jim Gray Computing Laws Beyond Moores Law The best way to predict the future is to invent it. --Alan Kay Gordon Bell Bay Area Research.

Documents

distributed computing

ubiquity of computing

p slide

new stores slide

new platforms

moores law results

shelf moores law

new class