CS267: Lecture 6, Feb 4 1999

Distributed Memory Computers (continued)

Data Parallel Architectures and Programming

Abstract

We finish our discussion of simple gravity algorithms and their implementation on distributed memory machines from last time.

Then we talk about data parallelism, including vector architectures, Fortran 90 and HPF (High Performance Fortran).

Lecture Notes

  • Power Point version
  • PDF version
  • Sharks and Fish code discussed in class, in CMF (no gravity)
    fish1.fcm is the main file
  • Sharks and Fish code discussed in class, in CMF (with gravity)
    fish2.fcm is the main file
  • Sharks and Fish code discussed in class, in Matlab (no gravity)
    fish1.m is the main file
  • Sharks and Fish code discussed in class, in Matlab (with gravity)
    fish2.m is the main file

    • Primary Readings

  • Notes from Lecture 4, from CS267, Spring 1996

    • Secondary Readings and References

  • Information about the Cray T90, a leading vector machine
  • Information about the SUN Visual Instruction Set, vector processing support for multimedia applications
  • Sharks and Fish, detailed programming examples of several simple physical simulations written in several programming models, including data parallelism (currently using CMF -- Connection Machine Fortran -- instead of HPF, but these are very similar).