Distributed Computing HOWTO

by Przemyslaw Kaminski <coolgenie@o2.pl>


This document describes most of the available distributed computing projects.

1. Introduction

2. Voluntary distributed computing projects

3. Commercial projects

4. Final thoughts


1. Introduction

1.1 What is the purpose of this document?

" Computer -- an electronic machine that performs high-speed mathematical or logical calculations and assembles, stores, correlates, or otherwise processes and prints data." ['The American Heritage Dictionary' (C) 1983]

Ever since PC's appeared and computers have become more popular to the 'common man' the idea of computers being the tool of scientists to do various processing of data has become distant. Today compuers are used as tools (word processing, spreadsheets, presentations) or entertainment centers (sound, video, 3-D games). The computer is no longer a complicated calculator but rather a technologically advanced toy.

The goal of this document if to change this stereotype. I want to make you more sensitive -- look at how much computing power is wasted when you do simple word processing on the brand new Pentium 4 with 2.4 GHz! Why not use that spare processor cycles to do something useful?

1.2 What is distributed computing?

" Distribute -- to divide and give out in portions; to hand out; deliver" ['The American Heritage Dictionary' (C) 1983]

Distributed computing is a simple idea: it makes long and complicated scientific computing a lot cheaper and faster (it depends on project's popularity). Instead of spending millions and millions of dollars on renting the top IBM ASCI White scientists prefer giving out a small piece of their work to each one of thousands of users all around the world participating in the project. The numbers will say it all: ASCI White is rated at 12.3 teraflops (1 teraflop = 1 trillion operations per second) while in 2001 the power of all computers gathered around the SETI@home project was... 20 teraflops (which equals to 200 years of constant work of a Pentium III 800 MHz processor). And that's only a fraction of all the power available (think of millions of users connected to the Internet at this time!)

What are you waiting for? They need you!

1.3 Can I make profits with distributed computing?

Well, don't expect much. This business is just getting started -- most of the projects are based on voluntary computing and only sometimes they provide you with a (rather small) chance of winning a prize.

1.4 What do I need to start?

I'll make it short -- if you've downloaded this text from the Internet then you can participate in a distributed computing project. Congratulations!


2. Voluntary distributed computing projects

Here are some non-commercial projects that you might be interested in. If you know other than the ones listed, e-mail me.

2.1 Up in the clouds [astronomy]

SETI@home (http://setiathome.ssl.berkeley.edu/)

This project is a major success. Started in 1999 it has now over 3 million of users all around the world in 170 countries. This project is only a part of the SETI (Search of Extraterrestial Intelligence) government program (" Independence Day" !!!) If you're a fantasy kind of guy then this is what you're looking for.

Advantages:

* very nice statistics

* create/join a team

* find the alien!

Disadvantages:

* still haven't found that alien...

* very serious server problems (always download some workunits ahead, 'cause you'll never know what'll happen...)

* one workunit takes 8 hours on an Athlon 900 MHz (although some people like big workunits)

* workunits are pretty big (about 350 KBytes in size) and really take a lot of time to download (that damn server...)

XPulsar (http://www.xpulsar.de)

2.2 Math problems

distributed.net (http://www.distributed.net)

This one's my favorite (join 'The Thirteenth Apostle'!!!). These guys crack the RC5 code to prove the government that the encryptino methods they use are inefficient (don't worry, this project is legal!) And you can win money! You'll get $1000 for finding the right key and another $1000 if you don't belong to a team (otherwise your team gets $1000).

But what I like the most about this project is the software. No other project has such configurable client/server software like distributed.net! If you want to see how bad (or good...) other projects are, I'd suggest downloading a distributed.net client and then comparing other clients to it. You'll always say that distributed.net is the best!

Advantages:

* the software rulez!

* create/join a team

* small, light workunits

* fast download -- no broken servers

* win money!

* if you don't like RC5, you can also compute OGR or CSC and DES

* the cow!!!

Disadvantages:

* it can be hard for computer dummies to configure

* little chance of winning the prize

Mersenne (http://www.mersenne.org/prime.htm)

2.3 Medical

The cancer project at United Devices (http://www.uniteddevices.com)

Folderol (http://www.folderol.org)

Folding@home (http://foldingathome.standford.edu)


3. Commercial projects

3.1 United Devices (http://www.uniteddevices.com)

3.2 Applied Meta (http://www.appliedmeta.com)

3.3 Data Synapse (http://www.datasynapse.com)

3.4 Distributed Science (http://www.distributedscience.com)

3.5 Parabon (http://www.parabon.com)

3.6 Porivo (http://www.porivo.com)

3.7 Entropia (http://www.entropia.com)


4. Final thoughts

If you have any questions or comments, e-mail me at <coolgenie@o2.pl>.