Table of Contents for CPS615-Introduction-Course,Driving Technology and HPCC Current Status and Futures

---

---

CPS 615 Lectures 1998 Fall Semester -- Introduction

• This Presentation is made up by selecting foils from modules or or foilsets which are sometimes specific to CPS615 and sometimes generic
• Some of these foils come from previous years courses and some are updated
  See 1997 Session of CPS 615 Basic Overview of Computational Science -- Simulation Track for last years complete course information

  ---

---

Start new Section:Overall Structure of Course

• Personnel Grading and Topics

1 Overview of Computational Science
2 Abstract of Computational Science Presentation

---

Technology Snapshot
3 Some Notes on Lecture Technology

---

What is Course
4 What is Computational Science ?
5 Synergy of Parallel Computing and Web Internetics as Unifying Principle
6 Basic CPS615 Contact Points
7 Course Organization
8 Material Covered in this Course
9 Structure of CPS615 - II
10 Basic Structure of Complete CPS615 Base Course on Computational Science Simulation Track -- III

---

What is Parallel Computing and Parallel Application
11 What are Parallel and Distributed Computing?
12 Why Parallel Computing?
13 Parallel Computing Technology Rationale
14 Motivating Applications

---

Performance Reality/Dreams
15 Performance of High End Machines Years 1940-2000
16 Performance of High End Machines Years 1980-2000
17 Peak Supercomputer Performance

---

Start new Section: Overview of HPCC Nationally

• National Program and Driving Applications

18 Some Comments on Simulation and HPCC

---

Start new Subsection:General Overview of Parallel Computing Technology
19 The Multicomputer: an Idealized Parallel Computer
20 Multicomputer Architecture
21 Multicomputer Cost Model
22 Sequential Memory Structure
23 Parallel Computer Memory Structure
24 Real Parallel Computers Architectures
25 Parallel Computers -- Classic Overview
26 Distributed Memory MIMD Multiprocessor
27 Distributed Memory Machines
28 Distributed Memory Machines -- Notes
29 Shared Memory MIMD Multiprocessor
30 Shared-Memory Machines
31 Shared-Memory Machines -- Notes
32 Distributed Shared Memory (DSM)
33 Distributed Shared Memory Machines
34 Workstation Clusters

---

Categories of Algorithms
35 Parallel Algorithms
36 Data Parallelism in Algorithms

---

Data Parallel application Examples:The Fundamental Reasons Why Parallel Computing is Easy

• In Principle
• Most Problems are naturally parallel and most can be naturally broken up
• Into parts for execution on separate Processors
• This is however fraught with technical difficulties

37 Some Illustrative Examples of Parallel Applications!
38 Concurrent Computation as a Mapping Problem -I
39 Concurrent Computation as a Mapping Problem - II
40 Concurrent Computation as a Mapping Problem - III
41 Finite Element Mesh From Nastran
(mesh only shown in upper half)
42 A Simple Equal Area Decomposition
43 Decomposition After Annealing
(one particularly good but nonoptimal decomposition)

---

Functional and Pleasingly Parallel Parallelism
44 Functional Parallelism in Algorithms
45 Structure(Architecture) of Applications - I
46 Structure(Architecture) of Applications - II
47 Multi Server Model for metaproblems
48 Multi-Server Gateway Tier
49 Pleasingly Parallel Algorithms

---

Parallel Computing Software
50 Parallel Languages
51 Data-Parallel Languages
52 Message-Passing Systems
53 A Simple Parallel Programming Model
54 Properties of Programming Model
55 Some Steps in Parallel Programming
56 Partitioning
57 Communication
58 Agglomeration
59 Mapping

---

Computer architectures and Technology Trends
60 What is Parallel Architecture?
61 Why Study Parallel Architecture as a computer scientist?
62 Why Study Architecture Today?
63 Inevitability of Parallel Computing
64 Application Trends
65 TPC-C (database transaction processing)
66 Summary of Application Trends
67 Technology Trends -- CPU's
68 General Technology Trends
69 Technology: A Closer Look
70 Clock Frequency Growth Rate
71 Transistor Count Growth Rate
72 Similar Story for Storage

---

Elementary Discussion of Parallel Computing

• Including Analogies with Society

  ---

73 Parallel Processing and Society
74 Concurrent Construction of a Wall
Using N = 8 Bricklayers
Decomposition by Vertical Sections
75 Quantitative Speed-Up Analysis for Construction of Hadrian's Wall
76 Amdahl's law for Real World Parallel Processing
77 Pipelining --Another Parallel Processing Strategy for Hadrian's Wall
78 Hadrian's Wall Illustrates that the Topology of Processor Must Include Topology of Problem
79 General Speed Up Analysis
80 Nature's Concurrent Computers
81 Comparison of Concurrent Processing in Society and Computing

---

More Complex Problem Issues in the Society Analogy
82 Comparison of The Complete Problem to the subproblems formed in domain decomposition
83 Hadrian's Wall Illustrating an
Irregular but Homogeneous Problem
84 Some Problems are Inhomogeneous Illustrated by:
An Inhomogeneous Hadrian Wall with Decoration
85 Global and Local Parallelism Illustrated by Hadrian's Wall
86 Parallel I/O Illustrated by
Concurrent Brick Delivery for Hadrian's Wall
Bandwidth of Trucks and Roads
Matches that of Masons

---

A Real Example
87 Example: Atmosphere Model
88 Atmosphere Model: Numerical Methods
89 Atmosphere Model: Partition
90 Atmosphere Model: Communication
91 Atmosphere Model: Agglomeration
92 Atmosphere Model: Mapping

---

Web and Java -- The Future
93 The HPCC Dilemma and its Solution
94 What is Commodity Software
95 The Computing Pyramid
96 Implications of the Computing Pyramid
97 The 3 Roles of Java
98 Why is Java Worth Looking at?
99 What is Java Grande?
100 Java and Parallelism?
101 "Pure" Java Model For Parallelism

---

See Java in HPCC resource
102 Java for Scientific Computing Resource

---

Pragmaticism
103 Pragmatic Computational Science August 1998

Click outside pointer rectangle to move pointer
Click on Pointer to Hide
Click on Pointer + ALT to toggle message hiding
Click on Pointer + CNTL to abolish pointer
Click on Pointer + Shift to cycle families
Click outside + Alt is Change Image
Click outside + Control is Double Size
Click outside + Shift is Halve Size

Right Mouse Down on Pointer Toggles Index
Shift Right Mouse aligns top with scrolled Page
While With Mouse Down on Current Pointer
h hides This Message while m restores
i Toggles Index Aligned with Page Top
j Toggles Index Aligned with Scrolled View Top
a Abolishes Pointer while CNTL-Click restores
f cycles through pointer families
c cycles through members of a family
u increases Size Up and d decreases Down
Mouse Up-Down between changes of
Pointer to process new option