Insight into some major trends in IT and its impact …The challenge of innovation driven curricula change. Insight into some major trends in IT and its impact on Informatics curricula

The challenge of innovation driven curricula change. Insight into some major trends in IT and its impact on

Informatics curricula

Prof. Dr. Mark HarrisAssociated Professor for Technology Entrepreneurship & Innovation

Intel® Director Higher Education & ResearchEurope, Middle East, Africa

© Prof. Dr. Mark HarrisAssociated Professor for Technology Entrepreneurship & Innovation


The Skills needed in ICT are changing !

With all due respect to other disciplines….

• If Newton were to resurrect he could continue to lecture Math where he left off….

• If Heisenberg were to resurrect he could continue to lecture Physics where he left off….

• If Turing were to resurrect he would find his whole compute model has disappeared and would have to sit in on the current lectures to catch up….

• The speed of change in ICT disciplines is enormous



3

The need for flexibility !

While the industry very much welcomes framework standards of skills at the highest possible level, there is also the need for much flexibility:

• Innovational needs

• Regional needs

• Industry needs

• Company needs



4

The need for flexibility !

• Innovational needs

– Technology evolves faster than curricula !

• Regional needs

– Countries have regional centers with different needs

• Industry needs

– Different industries have different needs

• Company needs

– Companies, based on their business, have different needs



5

The Next Challenge:

What to drop ?

What to incorporate ?

Here the 80/20 rule probably also applies



6

The Industry Model (?)

Second Cycle Degree

First Cycle Degree Fle

xib

ility

80 20



Future Thrusts in ICT

We see (at least) three major future Thrusts:

Visual Computing

High Performance Computing

Green (by) IT / CleanTech

7

Visual Computing at Intel The Next Frontier



Rich Interaction and ImmersionNew Content Spiral

Integrated WorldUser Generated

Social Networking

Users Collaborate & Play

Scenario Play

Virtual Teamroom

Users Create

World of Warcraft Avatar

Eiffel Tower in Google Earth

Users Explore and Learn

Qwaq Treefort Virtual Room Machinima Globalkids..org

Users Enhance the Actual World

West Nile Virus VisualizationVisualizing Real world information

Dust storm in Morocco



SOCIAL NETWORKINGConsumers increasingly use the Internet for Peering and Collaborating

USER-GENERATED CONTENTPrimary Role of the Internet Shifting From

Communications to Content

BROADBAND CONNECTIVITYSufficient bandwidth for interactive, online 3D experience

3D GRAPHICS & INTERACTIONConsumers interact visually in 3D, beyond the

Keyboard, Mouse

VIRTUAL ECONOMYReal and virtual economies have

been connected

It is now

technologically possible,

socially acceptable, and culturally appropriate

to interact with others

within immersive

3D online environments.

And It’s Happening

Now…

http://bp3.blogger.com/_4yuh0-Ww7FI/RaxAguFHRMI/AAAAAAAABgA/pSMuOSm5xfM/s1600-h/Snapshot_018.bmp

http://www.flickr.com/photos/16289690@N00/268345236/



Computing Evolution: A Collision Course

Battle For Control Of The Computing Platform

Fixed Function

Programmable

General Purpose

Multi-threading Multi-core Many Core

Throughput Performance

Programmability

SMTSmall Number

Of Traditional

Cores

Arrays of

Throughput

Cores

DX7

DX9

DX10

Future

CPU

GPU

CPU

• Evolving toward throughput computing

• Motivated by energy-efficient performance

GPU

• Evolving toward general-purpose computing

• Motivated by higher quality graphics and

GP-GPU usages



Larrabee: Block Diagram

Mem

ory

Contr

oller

Multi-Threaded

Wide SIMD

Multi-Threaded

Wide SIMD

Multi-Threaded

Wide SIMD

Multi-Threaded

Wide SIMD

D$I$D$I$

D$I$

Mem

ory

Contr

oller

Mem

ory

Contr

oller

D$I$

Multi-Threaded

Wide SIMD

D$I$

Multi-Threaded

Wide SIMD

D$I$

Multi-Threaded

Wide SIMD

D$I$

Multi-Threaded

Wide SIMD

L2 Cache

. . .

. . .

Fix

ed

Function

Textu

re

Logic

Syste

m

Inte

rface

Dis

pla

y

Inte

rface

Mem

ory

Contr

oller

•Multiple IA cores

–In-order, short pipeline

–Multi-thread support

•1024-bit ring bus

•Dedicated texture logic

•Supports virtual memory

•16-wide vector units

–Extended x86 ISA

•Fully coherent caches



Larrabee: CPU-like Software Stack

Application

Driver

Larrabee Hardware

Tools/LibrariesPerformanceAnalysis Tools

User ProgramsGraphics Apps

Larrabee Native Apps etc.

UtilitiesDriver Control

Panel

DirectX OpenGL

PCIe/Display Driver

Rendering Pipeline

API

Driver Executive (µOS)

DevelopmentEnvironment

CompilerDebugger

Larrabee Native C/C++

Larrabee Native App



Challenge of Parallel Programming

?

Irregular Patterns and Data Structures

Increasing Cores (2→64+ Cores)

Vector Instructions (4→8+ Wide)

Cache and Interconnect Latency

Scale to Multi-Core Today → Hard

Scale to Many-Core Tomorrow → Harder



Ct: A Throughput Programming Model

TVEC<F32> a(src1), b(src2);

TVEC<F32> c = a + b;

c.copyOut(dest);

1 1 0 00 1 0 1 0 1 0 00 0 1 1

1 1 0 00 1 0 1 0 1 0 00 0 1 1

+

Thread 4

0 0 1 1

0 0 1 1

+

Thread 3

0 1 0 0

0 1 0 0

+

Thread 2

0 0 0 1

0 0 0 1

+

Thread 1

1 1 0 1

1 1 0 1

+

Ct JIT Compiler:

Auto-vectorization, SSE,

AVX, LarrabeeCore 1

SIMD Unit

Core 2

SIMD Unit

Core 3

SIMD Unit

Core 4

SIMD Unit

Programmer Thinks Serially; Ct Exploits Parallelism

Ct Parallel Runtime: Auto-Scale to Increasing Cores

User Writes

Core Independent C++ Code

Key Data Abstraction is the Nested Vector

Supports Dense, Sparse, and Irregular Data



Speedup

0

10

20

30

40

50

BlackScholes

BinomialTree

Monte Carlo Convolution CollisionDetection

Ct- 8C

Ct-1C

SSE

Excellent Scale-Up



Daniel PohlResearch Scientist

Intel Corporation

Daniel Pohl

Intel Corp.http://www.qwrt.de/



Progress in Ray Tracing

50 Intel® Xeon™Processors

4 Frames per Second640x480

IDF 2004

Yorkfield(45nm Quadcore)

~90 frames per Second

768x768

Games Convention

2007 Dual-X5365

(total: 8 cores)

~90 frames per Second

1280x720

Fall IDF 2007



Ron Fedkiw



What does the future look like?



Visual Computing:Acquiring, Analyzing, Modeling and Synthesizing Visual Environments

Immersive

User Interface

Photorealistic

3D Rendering

3D

Audio and Video

Multi-Modal

Computing

With a Programmable, Ubiquitous, and Unified Architecture



High Performance Computing

22



23

The shift to massive computing parallelism in commodity devices

• For decades Moore‘s law has defined the increase of Hardware component complexity as well as performance. Now with Multi-Core and Manycore technology, Moore‗s Law has been transported into the Software world, with massive implications.

• Current typical programming skills are not enough to take advantage of the resulting performance boosts now available to everyone.



The good news: Moore‘s Law isn‘t over yet

65nm process

2005

30nm 20nm

45nm process(20 nm prototype)

2007


2009

15nm


2011

Source: Intel

Technology Node (nm) 90 65 45 32 22

Integration Capacity (BT) 2 4 8 16 32

… combined with advanced packaging, we get the

familiar transistor-doubling with each generation

These are projections only and may not be reflected in future products from Intel Corp.



25

The bad news: Single thread performance is falling off (ILP at point of diminishing returns?)Historic SPECint 2000 Performance

Source: published SPECInt data

The free lunch is over…Herb Sutter, MS



Next step, Manycore: a fundamental design change - converging from many directions

general-purpose GPU• Tesla (NVidia)• Cell (IBM/Sony/Toshiba)• Firestream (AMD)

general-purpose manycore• Tilera (MIT)• RAMP (Berkeley),• Terascale R&D (Intel)• Larrabee (Intel, 2009)

heterogeneous manycore• Fusion (AMD, announced)

multicore evolving to manycore• Sun UltraSPARC T2: 8 cores x 8 threads = 64• IBM z10: 20 CPU cores (+ 2 service cores) • Intel Nehalem: 8 cores X 2 threads = 16 (2008)



Software challenge for manycore:traditional CS programs hard to change

Curriculum• what material gets dropped?• new course or change to existing one?

Material hard to find• textbook? (parallelism dropped from CLSR 2nd ed!)• lecture material, demos, labs?

What is the right model/approach/language to teach?



Manycore programming models: active R&D(in contrast to SMP Multicore)

Conceptual Models• “View from Berkeley”: 13 dwarfs• design patterns; RMS (Intel)• Alternatives to threads• everything old is new again?

• Linda, MPI, functional lang, skeletons(templates, frameworks (cactus)), actors…

Extensions to current languages• X10 (IBM/DARPA)• Ct (Intel CTG)• OpenMP, TBB: continue to scale?

Explicitly parallel programming languages:• Erlang (Ericsson)• TStreams (Intel/MIT)• Cilk, Haskell, Charm++, Titanium, etc, etc



ABCPL

ACE

ACT++

Active messages

Adl

Adsmith

ADDAP

AFAPI

ALWAN

AM

AMDC

AppLeS

Amoeba

ARTS

Athapascan-0b

Aurora

Automap

bb_threads

Blaze

BSP

BlockComm

C*.

"C* in C

C**

CarlOS

Cashmere

C4

CC++

Chu

Charlotte

Charm

Charm++

Cid

Cilk

CM-Fortran

Converse

Code

COOL

CORRELATE

CPS

CRL

CSP

Cthreads

CUMULVS

DAGGER

DAPPLE

Data Parallel C

DC++

DCE++

DDD

DICE.

DIPC

DOLIB

DOME

DOSMOS.

DRL

DSM-Threads

Ease .

ECO

Eiffel

Eilean

Emerald

EPL

Excalibur

Express

Falcon

Filaments

FM

FLASH

The FORCE

Fork

Fortran-M

FX

GA

GAMMA

Glenda

GLU

GUARD

HAsL.

Haskell

HPC++

JAVAR.

HORUS

HPC

IMPACT

ISIS.

JAVAR

JADE

Java RMI

javaPG

JavaSpace

JIDL

Joyce

Khoros

Karma

KOAN/Fortran-S

LAM

Lilac

Linda

JADA

WWWinda

ISETL-Linda

ParLin

Eilean

P4-Linda

Glenda

POSYBL

Objective-Linda

LiPS

Locust

Lparx

Lucid

Maisie

Manifold

Mentat

Legion

Meta Chaos

Midway

Millipede

CparPar

Mirage

MpC

MOSIX

Modula-P

Modula-2*

Multipol

MPI

MPC++

Munin

Nano-Threads

NESL

NetClasses++

Nexus

Nimrod

NOW

Objective Linda

Occam

Omega

OpenMP

Orca

OOF90

P++

P3L

p4-Linda

Pablo

PADE

PADRE

Panda

Papers

AFAPI.

Para++

Paradigm

Parafrase2

Paralation

Parallel-C++

Parallaxis

ParC

ParLib++

ParLin

Parmacs

Parti

pC

pC++

PCN

PCP:

PH

PEACE

PCU

PET

PETSc

PENNY

Phosphorus

POET.

Polaris

POOMA

POOL-T

PRESTO

P-RIO

Prospero

Proteus

QPC++

PVM

PSI

PSDM

Quake

Quark

Quick Threads

Sage++

SCANDAL

SAM

pC++

SCHEDULE

SciTL

POET

SDDA.

SHMEM

SIMPLE

Sina

SISAL.

distributed smalltalk

SMI.

SONiC

Split-C.

SR

Sthreads

Strand.

SUIF.

Synergy

Telegrphos

SuperPascal

TCGMSG.

Threads.h++.

TreadMarks

TRAPPER

uC++

UNITY

UC

V

ViC*

Visifold V-NUS

VPE

Win32 threads

WinPar

WWWinda

XENOOPS

XPC

Zounds

ZPL

Third party names are the property of their owners.

Implementations: Explicitly parallel languagesBe careful what you wish for….




Software challenge for manycore:What is Intel doing about this?

Research fundingUPCRC at Berkeley & Illinois (jointly with MS)PPL at Stanford (with several others)

ToolsIntel Thread Checker, Thread ProfilerThread Building Blocks (open source)

University ProgramMulti-core content and trainingCurriculum development grants



UPCRC: A Research Partnership

AcademiaH/W & S/W

Companies

Leadership$$Guidance

Innovation

Trained ResearchersGovernment

State Govs

Matching

Funds

Objective: Breakthroughs in Parallel Programming

Intel + Microsoft

$20 Million over 5 years

Matching University Investments

University of Illinois: $8 mil match

UC Berkeley: $7 mil match



Cloud Computing

Cloud Computing Testbed – Announced July 29, 2008

• ~5 site distributed testbed, each >1000-4000 cores, open to academic research community, dedicated to supporting the development of an open source stack

• Create a technology stack and research community to create the next 100 Google’s and Amazon’s

Reuters, Forbes, Financial times, BusinessWeek online, BBC, New York Times, Dow Jones Newswires, TechCrunch, CNET, EE Times, InformationWeek, ComputerWorld, Times UK, San Jose Mercury News, Bloomberg, … July 2008

http://welcome.hp.com/country/us/en/welcome.html



Making Parallel Computing Pervasive

EnablingParallelComputing

Academic Research

UPCRCs

Software

Products

Multi-core

Education

TBB Open Sourced

STM-Enabled Compiler on Whatif.intel.com

Parallel Benchmarks at Princeton‘s PARSEC site

Joint HW/SW R&D program to

enable Intel products 3-7+ in future

Intel Tera-scale

Research

Academic research seeking disruptiveinnovations 7-10+years out

Community and

Experimental Tools

Wide array of leadingmulti-core SW

development tools & info available today

Multi-core Education Program

- 400+ Universities - 25,000+ students - 2008 Goal: Double this

Intel® Academic Community

Threading for Multi-core SW community

Multi-core books



Intel Academic Community –Academic Community Collaborating Worldwide

In 72 countries around the world ,

at more than 959 Universities1626 professors areteaching 69700+ new software professionalsMulti-core programming



35

What we are doing to helpHelp mainstream programmers to quickly develop quality code

ResearchTomorrow’sTechniques

TransactionalMemory

Main threadMain thread Speculative Speculative

threadthread

SpawnSpawn

CommitCommit

speculative speculative

resultsresults

CC

BBAA

Main threadMain thread Speculative Speculative

threadthread

SpawnSpawn

CommitCommit

speculative speculative

resultsresults

CCCC

BBBBAAAA

SpeculativeMulti-threading

EducateTomorrow’sExperts

• 2006 - Helped 45 universities addparallel programming courses

• 2007 400+ univ. 25000 students• 2008 reaching out to 1000+ univ

Intel Academic Community

ACCELERATE TRANSITIONTO PARALLEL PROGRAMMING

ProvideToolsToday

Architect, Thread, Debug & Tune

http://www.intel.com/software/products



Intel® Academic CommunityGet Involved!

Connect with peers & technologists

Access new technology curriculum

Lead global discussion forums

Share your knowledge!– Join the discussions in the

forums– Give your opinions in the blogs– Add your content to the Wiki– Comment on what you are

viewing

http://academiccommunity.intel.com




Intel University Program

• Courseware to drop into existing courses

• Continual technology updates

• Free Intel® Software Development Tools licenses for the classroom

• Wiki and blog for collaboration

• Forums for specific questions

academiccommunity.intel.com

..\..\..\Local Settings\Temporary Internet Files\OLK53\academiccommunity.intel.com

Green (by) IT

ENERGY EFFICIENT IT STARTS WITH THE MICROPROCESSOR

BUT DOESN´T END THERE



Intel‘s Long History of Environmental Focus

Intel’s Climate Awareness Timeline

Ahead of the Curve

Transparency and Disclosure Driving Company-wideContinuous Improvement

Public Environmental

Reporting

Public Reporting of Total Energy Use

First GlobalSector-wide

Climate Change Goal

Public Reporting of Climate Footprint

Global Energy Conservation

Goal

First ICT Company to join EU

Commission’s Sustainable

Energy Campaign

LargestPurchaserof Green

Power in US

19941996

1998

2000

2003

2006

2008

―Green‖ Fab 32 Comes Online

Led industry

agreement on PFC

reduction: world‘s 1st voluntary

GHG reduction

agreement

2007

Join US EPA Climate Leaders: Global

Climate Footprint Goal



Intel‘s 2007 Carbon Dioxide Footprint for Operations

Total Carbon Dioxide Emissions3 million metric tons

24% PFCs

5% Other Chemicals

71% Mfg Energy

Source: Internal data, combined with World Resources Institute protocol for converting to CO2 impact.

Other chemicals includes N2O, fluorinerts, refrigerants, and CO2 generated from VOCs.

Note: Supply Chain not included.



Energy

Water

Chemicals

Chemical Waste Solid Waste

Wastewater

Global Warming GasesAir Pollution

Logistics &

Transport

Reducing Environmental Impact of Operations



Water

Chemicals

Chemical Waste Solid Waste

Wastewater

Global Warming GasesAir Pollution

Logistics &

Transport

Intel’s Ocotillo Campus wins EPA Water Efficiency Leader Award

WaterIntel’s newest 45nm processors to go lead free and halogen free

Intel, regulators and neighbors join together to reduce waste and increase recycling

Solid waste and consumer recycling initiatives reduce E-waste

Packaging reductions of 16-40% decreased number of shipments and fuel consumed

Intel wins the California Clean Air Award for outstanding leadership

Intel pledges to reduce global GHG emissions by 30% per production unit

Reducing Environmental Impact of Operations

Intel Ireland certified to IS393 Energy Management Systems Standard

Intel pursues LEED certification

for fabrication plants and buildings

Over 3 million gallons of water is reclaimed each year using special collection systems

http://www.intel.com/multi-core/index.htm?iid=tech_as+multi_img

http://www.epa.gov/projectxl

http://www.epa.gov/stateply/index.html



Our Current Challenge: Reducing Absolute Energy Use

Projected Energy Use with 5% Normalized Reduction

20

07

20

08

20

09

20

10

20

11

20

12

20

13

20

14

20

15

20

16

20

17

20

18

20

19

20

20

Abs Energy Norm Energy



Source: Internal data, combined with World Resources Institute protocol for converting to CO2 impact.

Other chemicals includes N2O, fluorinerts, refrigerants, and CO2 generated from VOCs.

Note: Supply Chain not included.

Total Intel Carbon Dioxide Footprint for Operations Plus Products 2007 (based on impact of products shipped each year)

Carbon Dioxide Emissions (37.5 million metric tons)

Commute

Logistics

Operations 3.0

Products 33.8 million metric tons

Intel’s Biggest Impact Area for the

Environment

0.5



90nm300mm

130nm200mm

180nm200mm

250nm200mm

65nm300mm

Dual Core

45nm300mm

Reduction In Transistor Switching Power ~30%

Improvement In Transistor Density ~2x

Improvement In Transistor Switching Speed>20%

Relentless Pursuit of Moore‘s Law Opens the Door to Innovation

Industry‘s 1st 45nm High-K Process Technology

Source: Intel



Performance Over Time*Power Reduction Over Time*

~ 1 million fold reduction in energy/transistor over 30+ years

Energy Efficiency Positive Impact On Environment

Pentium® Pro Processor

Pentium® -II Processor

Pentium® -IIiProcessor

Pentium® 4 Processor

Pentium® 4 Processor EE

Pentium® -D Processor

Core™ 2 Duo Processor X6800

Core™ 2 Duo Extreme QX6700

i486DX2

i486

386

Pentium® Processor

1986 200819881990199219941996199820002002200420061

10

100

1000

10000

Single Core Moore’s Law

Improving Energy Efficiency of Products



Improve computing energy efficiency by 50%

• Collectively save $5.5 billion in energy costs

Reduce global CO2 emissions by 54 million tons per year

• That is the equivalent of…

The Climate Savers Computing Initiative

planting 25,000m2

(~65,000km2) of treeselimination of 20

coal plants from the planet

removal of 11 million automobiles



Climate Savers Computing Initiative Members

http://images.google.com/imgres?imgurl=http://www.chine-informations.com/images/upload2/World_Wildlife_Fund_Logo.gif&imgrefurl=http://www.chine-informations.com/mods/dossiers/index.php?lg=en&action=fiche&id=1351&h=300&w=300&sz=5&hl=en&start=3&sig2=GUGfmNIkgo7hQ778FmFjRg&um=1&tbnid=YBKQ4YhK2iShnM:&tbnh=116&tbnw=116&ei=_Y1lRtimFqL8gQPy6dXtBg&prev=/images?q=world+wildlife+fund&svnum=10&um=1&hl=en&safe=off

http://images.google.com/imgres?imgurl=http://inventiv.com/graphics/NEC_logo.jpg&imgrefurl=http://inventiv.com/partner.html&h=120&w=228&sz=4&hl=en&start=32&sig2=KJWXy4EkOJeqwHJCaYImBw&um=1&tbnid=s_Bfo-M2h3iLHM:&tbnh=57&tbnw=108&ei=ljxoRvzlGKjgggPhjPDLBA&prev=/images?q=nec+computers+logo&start=20&ndsp=20&svnum=10&um=1&hl=en&safe=off&sa=N

http://images.google.com/imgres?imgurl=http://www.datacap.com/captivate/fujitsu-redps.jpg&imgrefurl=http://www.datacap.biz/captivate/sponsors/&h=310&w=619&sz=17&hl=en&start=9&sig2=OBTBWBvRHrFffk7QwfTQLQ&um=1&tbnid=LFqxXmi8k-VlXM:&tbnh=68&tbnw=136&ei=rTxoRsrUMKGEggOFiuSzBA&prev=/images?q=fujitsu&svnum=10&um=1&hl=en&safe=off

http://images.google.com/imgres?imgurl=http://www.iam.unibe.ch/das06/images/hitachi.jpg&imgrefurl=http://www.iam.unibe.ch/das06/sponsors.html&h=177&w=567&sz=34&tbnid=ZmgzzZn5iPRHqM:&tbnh=47&tbnw=150&prev=/images?q=hitachi+logo&um=1&start=1&sa=X&oi=images&ct=image&cd=1

http://www.marvell.com/index.jsp

http://www.ebay.com/

http://www.eg-core.de/index.php?nav=1

http://www.ubuntu.com/

http://www.bull.com/

http://www.obc.co.jp/



Total Intel Carbon Dioxide Footprintas a Portion of Global Emissions

3%

97%



Source: “A Smarter Shade of Green,” ACEEE Report for the Technology CEO Council, 2008.

Automobiles

Passenger

Airlines

Agriculture

Steel

Manufacturing

Lighting

Computer

Systems

40%

121%

132%

167%

339%

2,857,000%

1978 2008Energy-efficiency

Improvement

14.3 miles per gallon of

gas

20.0 miles per gallon of

gas

22.8 revenuepassenger

milesper gallon

50.4 revenuepassenger

milesper gallon

0.63 units ofoutput per unit

of energy use

63 pounds of steel per MBtu

Incandescent light bulb— 13

lumens per watt

1,400 instructions

per second per watt

40,000,000Instructions per second per watt

Greater Role of ICT in Enabling Energy Efficiency

1.46 units ofoutput per unit

of energy use

167 pounds of steel per MBtu

Compact flourescent

bulb—57 lumens per

watt



3%

97%



Questions ?

52

Documents

Insight into some major trends in IT and its impact …The challenge of innovation driven curricula change. Insight into some major trends in IT and its impact on Informatics curricula