Hardware Architectures and
Parallel Programming Models
An Introduction

07.11.01


Start Lecture by clicking here


Content

Author: Rolf Rabenseifner, Michael Resch

  1. Hardware Architectures and Parallel Programming Models
  2.     Contents
  3. Motivation
  4.     Motivation
  5.     Abstract Model
  6.     Parallel Compiler
  7. Hardware Architectures
  8.     We need the compute power
  9.     We can not go on like this
  10.     Progress of microprocessors
  11.     Why not vector computers anymore
  12.     Evolution of supercomputers
  13.     Other aspects - even more important ones!
  14. Basic Architectural Concepts
  15.     Old and new concepts
  16.     Pipelining
  17.     Array - Processor (I)
  18.     Array - Processor (II)
  19.     Multiprocessor - shared memory
  20.     Multiprocessor
  21.     Multicomputer - distributed memory (I)
  22.     Multicomputer - distributed memory (II)
  23.     The concepts of Flynn
  24.     I heard that ........
  25. Network Topologies
  26.     Complete interconnect
  27.     Complete interconnect
  28.     Parameters for network topologies (I)
  29.     Ring topology (Clusters)
  30.     Hypercube (very popular in CS)
  31.     2D-Mesh or Torus (Paragon, T3E)
  32.     Switch (SX-4, SX-5)
  33.     Current developments
  34.     Hybrid architectures
  35.     Hitachi SR 8000-F1/112 (Rank 5 in TOP 500 / June 2000)
  36.     Earth Simulator Project ESRDC / GS 40 (NEC)
  37.     Hardware Architectures Summary
  38. Parallelization Strategies
  39.     Parallel Compiler
  40.     A Problem (I)
  41.     A Problem (II)
  42.     Parallelization strategies (1)
  43.     Parallelization strategies (2)
  44.     Speedup, Efficency, and Scalup
  45.     Parallelization problems
  46.     Parallelization Problems
  47.     Amdahls Law
  48.     Amdahls Law (double-logarithmic)
  49. Programming Models
  50.     Concepts of Programming Models
  51.     Concepts - Shared Memory Directives (I)
  52.     Concepts - Shared Memory Directives (II)
  53.     Concepts - Shared Memory Directives (III)
  54.     Concepts - Data Parallelism (I)
  55.     Concepts - Data Parallelism (II)
  56.     Concepts - Message Passing (I)
  57.     Concepts - Message Passing (II)
  58.     Concepts - Message Passing (III)
  59.     Implementations
  60.     Implementations and Architectures
  61. Other Concepts
  62.     SHMEM - Shared Memory Interface
  63.     DMP Language Extensions
  64.     Multi Level Parallelism (MLP)
  65.     Example: Parallel Efficiency of OVERFLOW/MLP
  66.     MLPlib
  67. Advantages and Challenges
  68.     Advantages and Challenges
  69.     Programming Models on Hardware Platforms
  70.     Comparing Hardware Platforms
  71.     Which Model is the Best for Me?
  72.     Summary of Comparison
  73. Future Directions
  74.     Application Program Processing
  75.     Will We Converge? No! Parallel Programming Models Survey Results
  76. Parallel Programming Models Summary

Back to the Parallel Programming Workshop Overview