dan reed [email protected] multicore and scalable computing...
TRANSCRIPT
www.hpcdan.org
Multicore and Scalable Computing StrategistManaging Director, Cloud Computing Futures
2
“What probability of successful return would you accept to be the first human to set foot on Mars?”
3
Looking back, in the public mindThere were few or no experiences with …
web sites, email, spam, phishing, computer viruses
e-commerce, digital photography or telephony
Cell phones were rare and expensive
A Sony Walkman was still cool
WiFi was almost unknown
Similarly, on the computing frontThe dot.com boom was still underway
Robot vacuum cleaners were still science fiction
A terabyte was a lot of data
A peak teraflop was nation-scale infrastructure
The future depends on the vision and context …
4
Your car drives and navigates for you… and also parks the car (already a feature on some cars)
Your sound system only plays music you love… because it knows about every song you‟ve ever heard
Your phone only rings when you want to answer… because it knows your emotional state
All your family memories are recorded automatically… via MEMS-based sensors and solid state storage
Your body calls an ambulance when you‟re ill… via implanted, biologically powered diagnostic sensors
Your DNA sample determines personalized treatment… because genotype-phenotype models are specific
Your office adjusts its behavior to your needs… because it is smarter than you are
5
Your every physical movement is tracked/logged… by embedded sensors on all human artifacts
Your neighbors know all the books you read… because your electronic financial identity was stolen
Your every call is monitored for content… by deep semantic analysis and logging
Pollutant rationing determines lifestyle… via social network analysis and predictive characterization
Your utilities fail due to a virus attack… because security was penetrated by a 10 year old
Your DNA sample/lifestyle determine health cost… because you are targeted as a high risk genotype/lifestyle
Cyberwar destroys U.S. financial institutions…because U.S. lacks ability to construct IT infrastructure
6
A century agoAverage life expectancy was less than 50 years
Mean annual income was only a few hundred dollars
No more than 10 percent of houses had a telephone
A miniscule number of cars were on the roads
We did getScotch tape and crossword puzzles
Canned beer and iced tea
Self-heating coffee
But, we never gotThe flying cars
The underwater cities
Those shiny plastimetal clothes
7
Some rules of thumb In the near term, we overestimate change
In the long term, we underestimate changes
Outside their field of expertiseExperts are often better at predictions
The contra-Delphi effect
Inventing the future is more successfulRecognize exponentials
Quantitative change brings qualitative change
Recognize multidisciplinary coupling
Technological and social changeDifferent rates with differing consequences
8
“Don’t leave Earth without it.”www.cubesatkit.com
• Attributes
– ~$10K-$40K construction cost and ~$50K launch cost
• Secondary payload on commercial launcher
– 10 cm cube (one liter) to 10x10x30 cm
• Industry standard PC-104 boards
• See showcase.netins.net/web/wallio/CubeSat.htm
9
1890-1945
Mechanical, relay
7 year doubling
1945-1985
Tube, transistor,..
2.3 year doubling
1985-2003
Microprocessor
1 year doubling
2006-presentMulticore
Exponentials
Chip transistor density: 2X in ~18 months
Graphics: 100X in three years
WAN bandwidth: 64X in two years
Storage: 7X in two years
0
0
1
1,000
1,000,000
1,000,000,000
1880 1900 1920 1940 1960 1980 2000
Doubled
every 7.5
years
Doubled
every 2.3
years
Doubles
every year
Operations per second/$
Source: Jim Gray
Microprocessor
Revolution
10
Altair 8800 (1975)2 MHz Intel 8080A
Where Microsoft began
IMSAI 8080 (1976)2 MHz Intel 8080A
256 bytes to 8 KB
BASIC
Osborne 1 (1981)First “portable” computer
Original IBM PC (1981)4.77 MHz Intel 8088
16 KB to 640 KB memory
IBM BASIC and PC-DOS
11 11
MegabyteA small novel
GigabyteA pickup truck filled with paper or a DVD
Terabyte: one thousand gigabytes – ~$300 todayThe text in one million books
Entire U.S. Library of Congress is ~ten terabytes of text
Petabyte: one thousand terabytes1-2 petabytes equals all academic research library holdings
Coming soon to a pocket near you!
Routinely generated annually by many scientific instruments
Exabyte: one thousand petabytes5 exabytes of words spoken in the history of humanity
Source: Hal Varian, UC-Berkeley
1956
1972
12
13
It seems reasonable to envision, for a time 10 or 15 years hence, a 'thinking center' that will incorporate the functions of present-day libraries together with anticipated advances in information storage and retrieval.
The picture readily enlarges itself into a network of such centers, connected to one another by wide-band communication lines and to individual users by leased-wire services. In such a system, the speed of the computers would be balanced, and the cost of the gigantic memories and the sophisticated programs would be divided by the number of users.
J.C.R. Licklider, 1960
14
December 8, 1993, C Section (Front Page)
John Markoff, “A Free and Simple Computer Link - NCSA's Mosaic Program”
15
16
Big Iron (post WW II)Vacuum tubes and campy science fiction movies
Mainframe („60s/‟70s)Spinning tapes and bad science fiction movies
Workstations („70s/‟80s)Spinning disks and Star Trek™
PCs („80s/‟90s)Spinning CDs and Jurassic Park™
Internet („90s)Spinning DVDs and Internet pet food companies
Implicit computing (21st century)MP3 players and The Matrix™
Embedded intelligence and adaptive, mobile context
number of cores/person infinity
17 17
18
“Businessmen go down with their businesses because they like the old way so well they cannot bring themselves to change. …Seldom does the cobbler take up with a new fangled way of soling shoes and seldom does the artisan willingly take up with new methods of his trade.”
Henry Ford
19
Fre
e L
unch F
or
Tra
ditio
nal S
oft
ware
24
GH
z1 C
ore
12
GH
z1
Co
re6 G
Hz
1 C
ore
Op
era
tio
ns p
er
se
co
nd
fo
r se
ria
l co
de
No Free Lunch For Traditional Software(Without highly concurrent software, it won‟t get any faster!)
Additional operations per second if code can take advantage of concurrency
3 GHz8 Cores
3 GHz4 Cores
3 GHz2 Cores
3 G
Hz
1 C
ore
Faster Processors
New Software Features
Slower Programs
20
Three pillarsTheory, experiment and computation
Computation enables us toInvestigate phenomena where
economics or constraints preclude experimentation
Evaluate complex models and manage massive data model processes across interdisciplinary boundaries
Transform business and engineering practices
21
1963
Hahn and Lindquist
IBM 7090
One Processor
Each 0.2 MF
3 Hours
1977
Eppley and Smarr
CDC 7600
One Processor
Each 35 MF
5 Hours
1999
Seidel and Suen, et al.
NCSA SGI Origin
256 Processors
Each 500 MF
40 Hours
300X 30,000X
1,800,000,000X
2001
Seidel et al
NCSA Pentium III
256 Processors
Each 1 GF
500,000 Hours total
plus 500,000 hours at NERSC
~200X
22
“In the last two decades advances in computing technology, from processing speed to network capacity and the Internet, have revolutionized the way scientists work.
From sequencing genomes to monitoring the Earth's climate, many recent scientific advances would not have been possible without a parallel increase in computing power - and with revolutionary technologies such as the quantum computer edging towards reality, what will the relationship between computing and science bring us over the next 15 years?”
http://research.microsoft.com/towards2020science
23
The Standard Model quantum theory
Electroweak and strong forces
Gravity and relativityno “Grand Unified Theory”
Gravity integration and rationale for mass
Dark matter and dark energyMost of the universe‟s mass is “invisible”
Expansion is accelerating
Wilkinson Microwave Anisotropy ProbeUniverse is ~13.7 billion years old
Computing insightsLSST, LHC and QCD
24
Probing the Standard ModelThe Higgs boson and mass
Online at CERN “soon”Six detectors (2 general purpose)
International data hierarchy10-20 PB/year
CMS
ATLAS
25
Iterative solution
Conjugant gradient solver
Memory intensive
Machines built for just this problem
MILC
QLA
(QCD linear
Algebra)
QMP
QCD Message
Passing)
QMC
(QCD Multicore
Interface)
QDP/QDP++
(QCD Data Parallel)
QIO
(QCD I/O API)
QOP
(Dirac Operators,
inverters, forces)
Uniform User Env
(Runtime,
accounting, grid)
QCD Physics Toolbox
(Shared Alg, building blocks,
Visualization, performance tools)
Workflow and
Data analysis tools
CPSFermiQCDChroma
Level 4
Level 3
Level 2
Level 1
LQCD
Applications
Lattice Dirac Operator
Source: DOE SciDAC
26
Key project from the astronomy decadal survey
Celestial cinematography3 gigapixel detector for wide field imaging
ScienceBeyond the standard model
Dark matter and dark energy
Observation targets
Near Earth object survey
Weak lensing of wide fields
Supernovae measurements
Features>30 TB of data/night
Entire sky every 3 nights
Target first light 2013
27
Genomics
Proteomics
Cell biochemistry
and structure
Cilia
Mucus
Airway/flow
28
Computing, physics, engineering and biologyControl theory, mathematical models, phase spaces
From biological cartoons to predictive models
Timescale (seconds)10
-1510
-910
-610
-310
0 103
10-12
109
Siz
e S
ca
le
Ato
ms
Bio
po
lym
ers
Geologic &
Evolutionary
Timescales10
6
Org
an
ism
s
Ab initio
quantum chemistry
First principles
molecular dynamics
Empirical force field
molecular dynamics
Enzyme
mechanisms
Protein
folding
Homology-based
protein modeling
Evolutionary
processesEcosystems
and
epidemiology
Cell signalingCell
s
100
103
106
100
103
106
100
103
106
100
103
106
Organ function
DNA
replication
Finite element
models
Electrostatic
continuum models
Discrete Automata
models
Source: DOE
29
30
31
Biological (static and dynamic)DNA sequence and polymorphisms (static)
Gene expression levels
Biomarkers (proteins, metabolites, physiological …)
EnvironmentalAir and pollutants, particulates
Bacterial and viral distributions
Food and liquids
Mobility and exercise
Sociodynamic (physical and virtual)Spatial dynamics
Context and interactions
Electronic infosphere
32
Dynamic adaptationComputing, data and bandwidth
Deep integration and software services
Edge devices and large centers
Context-aware informationThe right information at the right time
The cloud follows you everywhere
33
“Consider a future device for individual use, which is a sort of mechanized private file and library. It needs a name, and to coin one at random, “memex” will do. A memex is a device in which an individual stores all his books, records, and communications, and which is mechanized so that it may be consulted with exceeding speed and flexibility. It is an enlarged intimate supplement to his memory.”
Vannevar Bush
“As We May Think,” Atlantic Monthly, 1945
34
Physical Virtual
35
MetricsLight, XYZ acceleration
Temperature, infrared
Current useAlzheimer‟s research
Canberra, Australia
36
Change is acceleratingFive years is an eternity
Information friction is declining
University adaptation is notSome hysteresis is good …
… but, we risk irrelevance
Frame debate by innovating Approaches, applications
Collaborations, benefits
Innovation accrues from risksFailure to risk is itself a risk
0 25 50 100 125 150
Automobile
75
Years
20
50
100 TelephoneElectricity
Radio
Television
VCR
PC
Cellular
% P
en
etr
ati
on
Source: Council on Competitiveness
37
“Every college student should be able to answer the following question: What is the relation between science and the humanities, and how is it important for human welfare? Every public intellectual and political leader should be able to answer that as well. … Most of the issues that vex humanity daily … cannot be solved without integrating knowledge from the natural sciences with that of the social sciences and humanities.” E.O. Wilson
38
Scholarly
communicationsDomain-specific
services
Instant
messaging
Identity
Document store
blogs &
social
networking
Notification
Search
books
citations
Visualization and
analysis services
Storage/data
services
Compute
Services and
virtualization
Project
management
Reference
management
Knowledge
management
Knowledge
discovery
Source: Tony Hey
39
“What probability of successful return would you accept to be the first human to set foot on Mars?”
40
In which year of birth (1930 or 1970) would you have the higher probability of walking on the moon?
Buzz Aldrin
(1930)
Neil Armstrong
(1930)
Pete Conrad
(1930)
Alan Bean
(1932)
Jack Schmidt
(1935)
Gene Cernan
(1934)
Edgar Mitchell
(1930)
James Irwin
(1930)
David Scott
(1932)
John Young
(1930)
Charles Duke
(1935)
Alan Shepard
(1923)
41
Technological,Scientific &
Societal Trends
© 2008 Microsoft Corporation. All rights reserved. Microsoft, Windows, Windows Vista and other product names are or may be registered trademarks and/or trademarks in the U.S. and/or other countries.
The information herein is for informational purposes only and represents the current view of Microsoft Corporation as of the date of this presentation. Because Microsoft must respond to changing market conditions, it
should not be interpreted to be a commitment on the part of Microsoft, and Microsoft cannot guarantee the accuracy of any information provided after the date of this presentation.
MICROSOFT MAKES NO WARRANTIES, EXPRESS, IMPLIED OR STATUTORY, AS TO THE INFORMATION IN THIS PRESENTATION.