Supercomputing and Sciences

29
Supercomputing and Sciences Rong Ge Marquette University

description

Supercomputing and Sciences. Rong Ge Marquette University. Roadmap. Supercomputing in plain English Personal computers and limited capability Supercomputers for solving scientific problems Supercomputing and speed Supercomputing for high school students Why should HS students care - PowerPoint PPT Presentation

Transcript of Supercomputing and Sciences

Supercomputing and Sciences

Rong Ge

Marquette University

Supercomputing in plain English Personal computers and limited capability Supercomputers for solving scientific problems Supercomputing and speed

Supercomputing for high school students Why should HS students care Supercomputing for HS in the country

Roadmap

Personal Computer

Output device

Input device

Input device

Network cable

Opening the Box

Processor: control and ALU Memory Input Output

Like human organs

Five Classic Components

Processor: number cruncher Speed: 2GHz-4GHz? Duo core or quad core?

Memory: data storage 8GB?

These hardware parameters largely determine how fast a computer is.

Typical PC Configurations

Are Long to compute Need large quantity of memory large quantity of runs Are Time Critical

Not All Programs can Run on PC

Slide 8

Example 2: Fluid dynamics calculations (1000 1000 1000 lattice)109 lattice points 1000 FLOP/point 10 000 time steps = 1016 FLOP

Example 3: Monte Carlo simulation of nuclear reactor1011 particles to track (for 1000 escapes) 104 FLOP/particle = 1015 FLOP

Decentralized supercomputing ( from Mathworld News, 2006/4/7 ): Grid of tens of thousands networked computers discovers 230 402 457 – 1, the 43rd Mersenne prime, as the largest known prime (9 152 052 digits )

Example 1: Southern oceans heat Modeling (10-minute iterations)300 GFLOP per iteration 300 000 iterations per 6 yrs = 1016 FLOP

4096 E-W regions

1024

N-S

re

gion

s

12 layers

in depth

Exemplar Programs

Physics and Astrophysics Biophysics Geophysics and Earth imaging Medical Physics and Medicine Chemistry and Biochemistry Chemical and nuclear reactions Weather and climate Mechanical devices - from prosthetics to spacecraft Manufacturing processes

Traditional Scientific and Engineering Problems

Top 1 in June 2012 Speed: 1016 operations per second today Big: 4500 square feet

Supercomputers

Supercomputers in the PastSource: Jack Dongarra

Source: Supercomputing in Plain English: Overview by Neeman at OU

12

Parallelism for Speed

Less fish …

More fish!

Parallelism means doing multiple things at the same time: you can get more work done in the same time.

Jigsaw analogy Person: CPU Jigsaw pieces: data in memory

One person Serial computing, one hour

Two persons Parallel computing, about a half hour

Four persons A little more than a quarter hour

Eight persons ?

13

Diminishing Returns

Source: Supercomputing in Plain English: Overview by Neeman at OU

1000 jigsaw pieces

Two person, each having on his own table with half of the puzzle pieces

Two persons can work completely independently, without any contention for a shared resource.

BUT, they need• Same number of pieces first – workload

decomposition and balance• Communication, which is costly

Supercomputing in Plain English: OverviewTue Jan 25 2011

14

Distributed Parallelism & Overhead

Supercomputing in plain English Personal computers and limited capability Supercomputers for solving scientific problems Supercomputing and speed

Supercomputing for high school students Why should HS students care Supercomputing for HS in the country

Roadmap

Tomorrow’s PCs may be today’s supercomputers

During the past 10 years, the trends indicated by ever faster networks, distributed systems, and multi-processor computer architectures (even at the desktop level) clearly show that parallelism is the future of computing.

Why Should We or Our Students CareReason I

Slide 17

CPU Performance

The exponential growth of microprocessor performance, known as Moore’s Law, shown over the past two decades (extrapolated).

1990 1980 2000 2010 KIPS

MIPS

GIPS

TIPS

Pro

cess

or

perf

orm

anc

e

Calendar year

80286 68000

80386

80486 68040

Pentium

Pentium II R10000

1.6 / yr

Slide 18

CPU Speed Projection in 2001

From the 2001 edition of the roadmap [Alla02]

Calendar year 2001 2004 2007 2010 2013 2016

Halfpitch (nm) 140 90 65 45 32 22

Clock freq. (GHz) 2 4 7 12 20 30

Wiring levels 7 8 9 10 10 10

Power supply (V) 1.1 1.0 0.8 0.7 0.6 0.5

Max. power (W) 130 160 190 220 250 290

The Truth

Microprocessor speed stops increasingaround 2003 due to physical difficulties

20

Multiple, slow cores on a chip Intel

Up to 80 cores AMD

Integrated CPU and GPU cores (50+ cores) nVidia

Hundreds of GPU cores Parallel computing is required to achieve fast

execution for a single program

The Resulting Multicore Processors

1. Thousand years ago – experimental Science Description of natural phenomena

2. Last few hundred years – Theoretical Science Newton’s Laws, Maxwell’s Equation

3. Last few decades – Computational Science Simulation of complex phenomena

4. Today – Data intensive Science Scientists overwhelmed with data sets

Reason II – Scientific Approaches

Need to solve grand challenge problems with supercomputing Disaster preparedness Climate change Clean energy National security and defense

Reason III: The Burden of Next Generation Scientists

Particle Physics

Swine Flu – Pandemic Flu Simulation

NSF and DOE

National supercomputing centers NCSA at UIUC San Diego supercomputer center the National Center for Supercomputing Applications

Technical supercomputing conferences IEEE/ACM Supercomputing XSEDE conference

Industry Intel Brings Parallel Computing to High School

Supercomputing for HS Programs

Supercomputing Organizations

Marquette University Several computer clusters

Guest accounts available Condor pool Technical help

SeWhip: Southeast Wisconsin high performance computing

Local Resources

Thank you