Home       About       Products       Vault       Store       Media        Press
    
    Parallel Computing Manager
    
Getting Started

This web site is designed to accomodate a variety of users. Which kind of user are you?
  • Creating Parallel Code - Are you already familiar parallel computing? This is the place to get into creating new supercomputer-compatible code using clusters.

  • Porting Parallel Code - Do you have a working MPI code? Start here to port your code to and run it on an easy-to-use Mac cluster.

  • Administering a Mac Cluster - Are you interested in administering a Mac cluster? See these links to administrative tools and example utilities.

  • Application Developer - Are you interested in writing GUI-based applications for a cluster? Learn how to parallelize your application.


Download
Purchase
Features
Scripting
Visualization
Command-Line Interface
Web Interface
MPI (Message-Passing Interface)
Parallel Computing Tutorials
Documentation
Other Links
Cluster SDK for HPC
Consulting Services

Follow these tracks introducing you to the resources available on the DR web site that you need initially. We provide additional materials addressing topics you will find increasingly useful as your experience and needs grow.


Creating Parallel Code

If you are already familiar with MPI and want to get started with a Mac cluster:

Start with the recipe and a trial version to build and begin using a cluster yourself

  • Mac Cluster Recipe - follow these simple instructions to build your personal computing cluster!

  • PCM Download - download a trial PCM and see it for yourself

then you may start writing parallel code using the examples in the SDK:

as a starting point. We find the built-in diagnostics and visualization tools in MacMPI to be very useful for debugging:
  • Visualization - take advantage of the communications visualization window present in MacMPI_X

If you'd like to experiment with other parallel computing codes, you may examine these codes:

  • Parallel Knock - an exhibition of elemental message-passing in a simple parallel code

  • Parallel Adder - a tutorial on converting a single-processor code with independent work into a simple parallel code

  • Parallel Pascal's Triangle - a tutorial on parallelizing a propagation-style single-processor code, implementing a calculation requiring local communication

  • Tutorials - a list of other tutorials on parallelizing code


Porting Parallel Code

If you have an existing MPI code and want to run it on a Mac cluster:

Start with the recipe and a trial version to build and begin using a Mac cluster yourself:

  • Mac Cluster Recipe - follow these simple instructions to build your personal computing cluster!

  • PCM Download - download a trial version and see it for yourself

then you may pick apart the makefile and project file examples in the SDK:

You may either merge this data with your existing code or substitute your code into these examples. We find the built-in diagnostics and visualization tools in MacMPI to be very useful for debugging:
  • Visualization - take advantage of the communications visualization window present in MacMPI_X


Administering a Mac Cluster

If you want to administrate a Mac cluster:

after designing your cluster and setting it up:

you can use the built-in queuing system in PCM or begin customizing your cluster access, including custom queuing, using AppleScript and/or Unix scripts using:

  • Scripting - see examples of scripts that direct PCM to scan for nodes and launch jobs

  • Unix CLI - direct PCM from the Unix command line using this open-source utility

Advanced administrators can learn how to administer users and compute time quotas in PCM:
  • Manual - see the documentation for details


Application Developer

If you have an application you'd like to parallelize for a Mac cluster:

Use the recipe and a trial version of PCM to build a cluster for your development:

then you can experiment with writing parallel code using the examples available here:

or use them as a starting point. We find the built-in diagnostics and visualization tools in MacMPI to be very useful for debugging:
  • Visualization - take advantage of the communications visualization window present in MacMPI_X

You might want to examine the source code of a parallel application:

  • Parallel Fractal GPL - source code to a numerically-intensive graphical application of cluster computing




© Copyright 2001-2006 Dauger Research, Inc. All rights reserved. PO Box 3074, Huntington Beach, CA 92605 USA