High Performance Parallel Programming

1
High Performance Parallel Programming

• Dirk van der Knijff
• Advanced Research Computing
• Information Division

2

• Lecture 1 Course outline and tutorial
  introduction

3

• Course Outline
• Introduction
• Architectures
• Programming
• Algorithms
• Case Studies

4

• Assessment
• Exam - 40 at end of semester
• Projects
• Introduction - 10 due mid-semester
• Main project - 50 due end of semester.

5

• Introduction
• During the rest of this lecture I want to give
  you a basic overview of High Performance
  programming and Parallel programming while I
  explore your knowledge in that area so that I can
  adjust the amount of introductory time to give to
  various topics.As such it will not be examined
  but hopefully will heighten your interest in the
  topic.

6

• Questions
• What is a Supercomputer?
• Why use one?
• Problems
• Scalability
• Portability
• Reliability
• Solutions???

7

• Quiz
• RISC

8

• Quiz
• RISC
• NUMA

9

• Quiz
• RISC
• NUMA
• CRAY

10

• Quiz
• RISC
• NUMA
• CRAY
• NOW

11

• Quiz
• RISC
• NUMA
• CRAY
• NOW
• COTS

12

• Quiz
• RISC
• NUMA
• CRAY
• NOW
• COTS
• SMP

13

• Quiz
• RISC
• NUMA
• CRAY
• NOW
• COTS
• SMP
• MIMD

14

• Quiz
• RISC
• NUMA
• CRAY
• NOW
• COTS
• SMP
• MIMD
• ASCI

15

• Supercomputing is old-fashioned computing
• You always need a bigger computer.
• Transaction processing is easy?
• What about SETI_at_home?
• Economics 1. You cant sell cycles!
• Economics 2. Programmers cost more
  than computers!

16

• Questions?

17

• Tomorrow - History

18

• Scientific Computing Characteristics
• Deals with BIG problems
• Deals with single programs
• System size is determined by smallest machine
  which can run program or largest machine
  available, whichever is smaller
• i.e. no cheating
• Match problem to machine (if possible)

19

• A Grand Challenge is a fundamental problem in
  science or engineering, with broad applications,
  whose solution would be enabled by the
  application of high performance computing
  resources that could become available in the near
  future.

20

• Computational Fluid Dynamics
• design of hypersonic aircraft
• design of efficient automobile bodies
• weather forecasting
• climate modelling
• efficient oil recovery

21

• Electronic structure calculations for the design
  of new materials such as
• chemical catalysts
• immunological agents
• superconductors
• Genetic Sequencing

22

• Plasma Dynamics
• fusion energy technology
• military technology
• Calculations to understand the fundamental nature
  of matter
• Quantum Chromodynamics
• condensed matter theory

23

• Symbolic computations including
• speech recognition
• computer vision
• natural language understanding
• automated reasoning
• Tools for design, manufacturing and simulation of
  complex systems

24

• Some definitions
• The fastest computer available today
• Any computer built by Seymour Cray
• A machine for turning a cpu-bound problem into an
  IO-bound problem

25

26

27

28