Defining Grand Challenges

National Symposium on High Performance Computing

Sponsored
by
Centre for Development of Advanced Computing , Pune
Bengal Engineering & Science University, Shibpur
4th March,2010

Pradosh K Roy FIETE

April 13, 2010 CDAC Symposium High 1

Performance Computing
 Statutory Declarations
• Pentium ® , Xeon ® are the registered trademarks of Intel ® Corporation.
• IBM™ Power is the registered trademark of IBM Corp.
• BSD ® is the registered trademark of UUnet Technologies Inc.
• UNIX ® is the registered trademark of Open Group.
• Mac ® is the registered trademark of Apple Corp.
• Linux® is registered at Free Software Foundation [FSF]
• Windows ® is the registered trademark of Microsoft Corporation.
• Myrinet ® and Myricom ® are registered trademarks of Myricom Inc.
• Infiniband is developed by Compaq ®, IBM ®, HP ®, Intel ®, Microsoft ® ,Sun ®
• ConnetX and InfiniHost are the registered trademarks of Mellanox Technologies Ltd.
• Portable Batch System [PBS] is developed at NAS,NASA.
• NAS Parallel Benchmark Suite [NPB] is developed at NASA Ames Research Centre, Calif.
• MPI is developed at Argonne National Laboratory, University of Chicago.
• PVM is developed by Oak Ridge National Laboratory
• OpenMP™ is the registered trademark of OpenMP Organization.
• CONDOR is developed at University of Wisconsin-Madison,Wisconsin
• Globus Tool Kit was developed by Team of Computer Scientists from ANL/EPCC/NCSA/USC is now
available under General Public License

April 13, 2010 CDAC Symposium High 2

Performance Computing
 Scheme Of Presentation

• Defining Grand Challenges & Nature of the Physical World

• Algorithms & Complexity [P#NP]
• High Performance Computing : Hardware/Software Interface
• Mark Baker & IEEE White Paper on Cluster Computing [2001]
• Parallel Programming Environments [ MPI ,PVM ,OpenMP]
• High Precision Computation
• National Policy for Supercomputing : NSF0728 Report
• Feynman & Theory Quantum Computing .
• Conclusion

April 13, 2010 CDAC Symposium High 3

Performance Computing
Grand Challenges : 21st Century Definition

A Grand Challenge is a long term science , engineering or

societal advance whose realization requires innovative
breakthrough in information technology research and
development and which will help address
Our Country’s [i.e. USA] Priorities

- Networking & Information Technology Research and

Development [NITRD] [US Govt. Agency]

April 13, 2010 CDAC Symposium High 4

Performance Computing
 Philosophy of HPC : Grand Challenges from
700BC to Modern Days
Grand Challenges in Ancient Civilizations

Plane Geometry
Astronomy
Shape and Radius of Earth [Erastosthenes 240BC]
Distance between Sun and Earth
Solar and Lunar Eclipses
Lunar Calendar
Navigation
Trigonometric Tables
Surya Siddhanta [700BC ?]
Almagest [110AD] , Siddhanta Siromani [1120 AD]
Number System
Babylonian [Sexagesimal][2000 BC]
Hindu-Arabic [Decimal][? 700BC]
Ionian
Arithmetic
3D Geometry of Pyramids
Algebra
Metals & Alloys
Medicinal Utility of Plants
Babylonian Plimpton 322

April 13, 2010 CDAC Symposium High 5

Performance Computing
 Hindu Trigonometry 700 BC(?)

Sine of 3¾°=225/3438 [0.065445026]

[Actual Value is 0.065403129]

April 13, 2010 CDAC Symposium High 6

Performance Computing
 Our Changing Vision of the Universe
……Keppler , Galileo, Newton, Einstein…….

April 13, 2010 CDAC Symposium High 7

Performance Computing
 Nature of Physical World
2500BC Babylonian/Hindu/Egyptian/Mayan Civilizations] [Astronomy]
500BC : Pythagorus [ Geometry]
450BC : Democritus / Plato
300BC-200BC : Aristarchus / Erastosthenes/Euclid [Geometry/Trigonometry/]
100BC- AD : Antikythera Machine
0AD-100 AD : Heron of Alexandria [Gear Trains]
100AD – 200AD : Ptolemy [Astronomy/Geography]
400AD-500 AD : Destruction of Alexandrian Library [Dark Age]
Da Vinci/[Italian Renaissance][Art/Science/]
500AD-1500 AD : Poignant Lost Opportunity for Human Species [Carl Sagan]
1500AD- 1600 AD : Copernicus/ Tyco Brahe/Keppler/Galileo [ Solar system l/Dynamics]
1600AD-1700AD : Isaac Newton/Huygens/Leibnitz [ Physical Science]
1700 AD- 1900 AD : Michael Faraday/Maxwell/ Hertz/ J C Bose[ Physical Science]
1900AD – 1930 AD : Einstein /Max Planck/Bohr/S N Bose/Raman
/Saha/ Heisenberg/Schroedinger /Pauli/Fermi
Lise Meitner/Paul Dirac/ Lev Landau/Feynman[ Physical Sciences]
1945 AD – 1995 AD : Astrophysics/Quantum Electrodynamics/Particle Physics//BioSciences
1995 AD- Present Day : Data Analysis/Visualization/
Simulation of Macro Phenomena/Bioinformatics/Nonlinear Phenomena &c
2025AD (?) - ? : Quantum Computers & NP Problems

April 13, 2010 CDAC Symposium High 8

April 13, 2010 Performance Computing
 Philosophy of HPC : 20th/21st Century Grand Challenges

Year System Speed System Memory Grand Challenge

1980 100 MFLOPS 10 MB 48 Hour Weather Modeling

1990 1 GFLOPS 5 GB -10 GB + 3D Plasma Modeling/72 Hour Weather

1993 100 GFLPOS >10GB + Pharmaceutical Design/Structural Biology

Geophysical Imaging/Reservoir Simulation

1995 1TFLOPS 103GB + Global Climate Change/CFD/Ocean

Circulation/Semiconductor Modeling/
Computer Vision/ Bioinformatics/
Vehicle Dynamics

2004* 106 GB + NASA Science Mission Directorate

Solar System Science
Origin of Life
Big Bang and Dark Energy …..

> 2010 1 PFLOPS 109 GB

•NASA Institute of Advanced Concepts

http://www.niac.usra.edu

April 13, 2010 CDAC Symposium High 9

Performance Computing
 Philosophy of HPC : 20th/21st Century Grand Challenges

Year System Agency Grand Challenge

Requirements
1995 1 TFLOP NSF Aerospace
Molecular Biology
Space Science
Environmental Modeling
DOE Energy
Environmental Monitoring & Prediction
Computational Chemistry
NASA Galaxy Formation
Convective Turbulence
Heliospheric Dynamics
NIH Molecular Biology
Biomedical Imaging
NIST Product Design & Process Optimization

NOAA Weather Forecasting

•Grand Challenges NITRD Govt of US
Third Printing – November,2006

April 13, 2010 CDAC Symposium High 10

Performance Computing
 Algorithms & Complexity
The Classical Theory

O(2n) ~ 366 Centuries @1979 Processor

Comparison n**2 & nlgn [FFT]
O(3n) ~1.31 x 1013 Centuries@ 1979 Processor
For n = 60

180000
160000
140000
120000
100000

n2, nlgn
80000
60000
40000
20000
0
-20000 n

Reference 11
1. Dasgupta, Papadimitriou, Vazirani : Algorithms
2. Abhiram Ranade ://www.youtube.com
April 13, 2010 CDAC Symposium High 11
[Design and Analysis of Algorithms]
Performance Computing
 Intel Microprocessor Performance

Computer Architecture 7th Edition : William Stallings

April 13, 2010 CDAC Symposium High 12

Performance Computing
 Supercomputer Architecture
1965 - 2000

April 13, 2010 CDAC Symposium High 13

Performance Computing
 HPC : Hardware/ Software Interface

IEEE TFCC Cluster Computing White Paper [2000]

April 13, 2010 CDAC Symposium High 14

Performance Computing
 HPC : Hardware/ Software Interface

IEEE TFCC Cluster Computing White Paper [2000]

April 13, 2010 CDAC Symposium High 15

Performance Computing
 • Obvious huge interest in Clusters, seemed natural to
set up a focussed group in this area.
• A Cluster Computing Task Force was proposed to the
IEEE CS, approved and started operating in February
1999 – been going just over 2 years.
• Task Forces:
– Expected to have a finite life (<3 years), longer
generally not appropriate.
– Either increase scope of activities or fade and die.
– TFCC will submit an application to the CS become a
TC later this year.
April 13, 2010 CDAC Symposium High 16
Performance Computing
 HPC Trends : Processor Family

Intel EM64T

Intel IA 32

http://www.top500.org

April 13, 2010 CDAC Symposium High 17

Performance Computing
 HPC Trends : Operating Systems

Linux

UNIX

April 13, 2010 CDAC Symposium High 18

Performance Computing
 Parallel Environment [GPL]
High level Languages : FORTRAN 77 , C , C++
[e.g. f77,gcc compilers in Linux]
Parallel Programming Environments

MPI [Message Passing Interface][ Arogonne National Lab]

[de facto standard for DMM/DSM Architectures]
PVM [Parallel Virtual Machine] [Oak Ridge National Lab]
Open MP [ For SMP Architectures]

Numerical Libraries
Numerical Recipes : The Art of Scientific Computing
W H Press , Saul Teukolsky , Vellering, Flannery [CUP ,1986]
IMSL [Visual Numerics Inc.],
NAG [Numerical Algorithm Group]

Cluster Management System

April 13, 2010 CDAC Symposium High 19

Performance Computing
 NASA Goddard Space Flight Centre
The Cluster Supercomputer

1999 The Stone Souper ORNL

April 13, 2010 CDAC Symposium High 20

Performance Computing IBM Blue Gene
 Simulation of Black Hole Merger
An Example of Simulation

Merger of Black Holes and Ripples in Space Time

Courtesy : National Centre for Supercomputing Applications , University of Illinois.

April 13, 2010 CDAC Symposium High 21

Performance Computing
 Simulation of Net Ecosystem Exchange at
SciDAC ,ORNL ,2009

As the Sun rises over Eastern Europe, the instantaneous net ecosystem exchange (NEE) of CO2 is shown in
the Eastern Hemisphere. Strong uptake is shown in green-to-white colors and is strongest in the tropics.
A net release of CO2 to the atmosphere is shown in red-to-white colors and is strongest over the Congo,
where the Sun is not shining. This image was produced from a C-LAMP simulation performed as part of a
SciDAC-2 project using NCCS supercomputers. ©DOE, SciDac, ORNL, USA

April 13, 2010 CDAC Symposium High 22

Performance Computing
 High Precision Computing
REFER TO THE WORK DONE BY DAVID BAILEY AND HIS TEAM IN
LAWRENCE BERKELEY NATIONAL LABORATORY AT
http://crd.lbl.gov/~dhbailey/dhbpapers
And/Or Jonathan M Borwein , David Bailey , Mathematics by
Experiment: Plausible Reasoning in 21st Century , AK Peters,MA,2004

S EEEEEEEEEEE FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF 0 1 11 12 63
The value V represented by the word may be determined as follows:
If E=2047 and F is nonzero, then V=NaN ("Not a number")
If E=2047 and F is zero and S is 1, then V=-Infinity
If E=2047 and F is zero and S is 0, then V=Infinity
If 0<E<2047 then V=(-1)**S * 2 ** (E-1023) * (1.F) where "1.F" is intended to represent the binary number created by
prefixing F with an implicit leading 1 and a binary point.
If E=0 and F is nonzero, then V=(-1)**S * 2 ** (-1022) * (0.F) These are "unnormalized" values.
If E=0 and F is zero and S is 1, then V=-0
If E=0 and F is zero and S is 0, then V=0

April 13, 2010 CDAC Symposium High 23

Performance Computing
 Necessity for
National Supercomputing Policy

NSF VISION FOR 2010 [ vide NSF0728]

• HPC System and Services for
Peta-scale Scientific Computing
2. At Least one System capable of sustaining 1015 FLOPS
3. Development and Maintenance of Performance Tools ,
System Software and Fundamental Algorithms.
4. Development and Maintenance of Portable Scalable
Application Software and Standards[ e.g CDAC ONAMA].
5. Development of National Digital Data Framework.
6. Development of Networked Resources and
Virtual Organizations.
7. Using Cyber-infrastructure for Enhancing Learning.

April 13, 2010 CDAC Symposium High 24

Performance Computing
 HPC : A Business Perspective

• High Performance Application Software

- Seismic Imaging , Reservoir Simulation
- Weather Modeling & Forecasting
- Semiconductor Modeling
- Aerospace Engineering & Simulation
- Data Mining
- Visualization &c.
• High Performance Hardware Architecture
• High Performance System Software Development /Services
• High Availability Network : Design & Services
• Data Centre Architecture : Design & Services
• ad infinitum …….

April 13, 2010 CDAC Symposium High 25

Performance Computing
 Future of Supercomputing
Feynman & Quantum Computing

Prime Factorization [Peter W Shor, 1987]

Classical Algorithm O(n) = exp[2 (lg n) 1/3 (lg lg n)2/3]

Shor’s Algorithm O(n)[1987] = (lg n)3
[arxiv:quant-ph/9508027v2/ 25th Jan 1996]

Searching Unsorted Database [L K Grover 1996],

Classical Algorithm O(n) ≈ n/2 Operations
Grover’s Algorithm O(n) = √n Operation
[arxiv.org:quant-ph/0109116 22nd September,2001]

April 13, 2010 CDAC Symposium High 26

Performance Computing
 Conclusion
• The Known is finite , the unknown infinite ,
intellectually we stand on an islet in the midst of an
illimitable ocean of inexplicability. Our business in
every generation is to reclaim a little more land
- T H Huxley

April 13, 2010 CDAC Symposium High 27

Performance Computing
 Selected References
1. The Sleepwalkers, Arthur Koestler , Penguin Books,1969
2. Act of Creation, Arthur Koestler, Penguin Books, 1964
3. Cosmos , Carl Sagan, Ballantine Books,1974
4. Surya Siddhánta , Swami Vijnanananda, Sanskrit Book Depot,Calcutta,2005
5. Atomic Physics & Human Knowledge, Niels Bohr, Braunschweig, 1985
6. The Science of Leonardo , Fritjof Capra, Anchor Books,2007
7. Feynman Lectures on Physics , Vol 3, Addission Wesley Inc.,1964
8. Feynman Lectures on Computation , Perseus Books ,1996
9. Using MPI , W Gropp,E.Lusk , A.Skjellum, The MIT Press , 1994
10. Computer Architecture : Hardware Software Interface, Hennessey/Patterson ,
11. Parallel Computer Architecture : David Cueller, Jaswinder Singh, Anoop Gupta
12. Great Ideas in Theoretical Computer Science, 6.089, MIT Open Courseware,2008
…………………………….

THANKS
April 13, 2010 CDAC Symposium High 28
Performance Computing