(updated 28. Aug 2004 and 31. Mar 2005, mainly about BOINC) 

New own BOINC/SETI client download page

Some years ago some people had an idea: if their own resources of computing power are insufficient, but the problems of a more general interest and well suited for dividing up into more or less independent files, which can be calculated seperately, why not invite computer users all over the world to give them some things to calculate, while their computer are turned on but idle? First distributed net took up this challange, last the SETI@home effort made the biggest impact and became best known. Generally the way of working of the client is different and depending of the operating system used. The two most important now are handled this way: on the always perfectly suited UNIX and LINUX machines the processes are started as nice ones, therefore not taking away CPU time from more important processes on the system, but taking every not by these others needed CPU cycle for their own purpose. For the vastly inferior MS Windows a sidestepping of the inappropriate support was invented: as screensaver or permanently background task; but the last is only acceptable for fast computers and is always costing performance for other programs. However, these projects are classic number cruncher tasks and at best you take as many as fast computers you can and are allowed to for the chosen project...

In the following sections I provide some general informations about the more important and widespread projects of distributed computing with a short --- of course slightly subjective --- evaluation of their importance. If you like the general idea, I recommend strongly to participate in one of them; everybody should be able to pick one of this list, which is to his/her like. If you know any other major ones, please use my email address at the bottom of the page.

BOINC @ Berkeley

BOINC Distributed Computing Framework

It is an open source client mechanism, which can control several BOINC projects at once: you can for example install this one, create accounts with several projects and let participate some machines in one of them, others in another, third in both together, and give priorities and resource limits for all. The main BOINC client is a master process, which controls the several project control processes, which control their own project client programs. Sounds complicated, but is rather flexible and not really complicated at all. Both highest priorities projects of mine are now availabel as BOINC version, though they have non-BOINC predecessors.

SETI@home

SETI BOINC: the new style SETI@home
Berkeley site of SETI@home classic  further details

Aim: detection of synthetic radio signals from other worlds in the universe, of course worlds with intelligent, technical sufficient developed beings.

This is --- as mentioned above --- at now the most users and CPU time accumulating project of this type. It's possible, to build even groups of users to give the participation more structure.

Meaning: it's a matter of belief, if you think that this effort makes sense or not. Despite of it's low probability of success in it's four to five year duration (not to mention the BOINC related extension into future) even in the view of the most participants and creators, the consequences of a success would be tremendous and therefore I think this is not only the most popular one now, but also the or at least one of the most important.

Important Remark: since 14. May 2004 there is the transition running of old style SETI@home classic to new style SETI/BOINC, a final transition date is not announced yet at time of writing (28. August 2004), so changing is at now not necessary, but possible. New participants should go for the BOINC version of course.

Climatology Simulation

Anchor Site

Aim: to calculate as many possible climatologic changes in the future as possible for a reliable conclusion about the situation and eventual solutions.

Update: the launch is now for the first time scheduled, the aim was August 2002, the true public release was then launched only at 12. September 2003. The major drawback is, that against the original planning at first no GNU/Linux version was available, but only M$ Windoze versions --- but this changed on 26. August 2004: therefore I'm now able to participate...

Latest News: finally the climate prediction distributed computing campaign switched over to BOINC too and released versions for Intel-GNU/Linux and Mac OS X (10.3 and later), and I have switched also some machines fulfilling the high hardware requirements and running one of the so far only two supported Unix flavors to this project.

The BOINC client runs as command line utiltiy in Unix systems and a separate visualization client can be launched with a script named viz (in case you have several models running on different CPUs even with a parameter for model selection), but it costs about the same CPU power than the cruncher client, so I suggest for best performance to avoid running it too or at least only occasionally for your information.

Meaning: IMHO it is the most important of the distributed projects I know so far. Therefore personally I have switched sufficient fast machines to it from the SETI project, because there is an issue with the not very satisfying chance of finding anything worth mentioning it (a few percent at most with optimistic guesses) in SETI projects nowadays. --- Keep in mind, that we need urgently to understand in detail, how we are shifting the equilibrium of Earths atmosphere into an unsound area, to prevent the worst and eventually to work even active against it.

Distributed Net

Joint Homepage(s)  further details about the Golomb part

Aim: to tackle diverse topics via Internet number crunching: DES decoding as security problem, optimum Golomb ruler search and a few others.

This is a bundle of tasks, which is performed by the very same client. But you are able to choose one or more (or of course the default: all) of the projects and ignore the others.

Meaning: especially the Golomb ruler search can deliver valuable results, because these optimum solutions can improve interferometry systems and similar things. The DES decoding is a type of checking, how (un-)safe this encoding is for our available computing environment. For more informations about other parts please resort to above homepage. Importance is ranked as medium one by me.

German flagDeutsches Projekt: Xpulsar at home

Uni Tübingen

Projekt ist abgeschlossen! Keine Fortsetzung absehbar! Ziel und Methode: Ein Java-Applet für den Internet-Browser: eine etwas andere Methode als sonst; es kann speziell für LINUX allerdings auch als kompiliertes Program in JRE gestartet werden. Es geht um die komplizierte Berechnung der Photonenbewegung in einer Riemannschen Metrik (allgemeine Relativitätstheorie). Auch dieses Projekt beinhaltet eine Statistik der User.

Bedeutung: ein wissenschaftlich interessantes Projekt, dessen Ertrag allerdings zunächst reine Grundlagenforschung darstellt --- auch wenn man nie weiss, ob eines Tages nicht irgendein Nutzen daraus gezogen werden kann. Mittlere bis kleinere Bedeutung.

Mersenne Prime Search

GIMPS homepage  more details

Aim: this is a record breaking effort: the hunt for the biggest primes. Due to todays known algorithms the Mersenne numbers can be tested most effectively, therefore the biggest known primes are all of this kind.

There are different types of calculations going on: random tests in new high areas of exponents and systematic evaluation of lower ranges; all found numbers are then double-checked and the range below fully covered, to ensure the order of found numbers. Four were found so far by the project, and most experts agree, that this task will deliver always further ones...

Meaning: this is just for the interest of number theory fans like me, I guess. Even for factoring purposes this task is meaningless; therefore it's only here because of my personal preferences...

Further Projects:

Google sorted overview about these projects and another short overview list

The only --- and very good! --- comprehensive source(s) about this way of Internet usage, which I could find so far. It will always pay off, to look sometimes into this overview list; it is not too long, but long enough for checking what's else going on in this interesting field of global collaboration!


 

back to main

remarks, questions, proposals etc. to: stefan.urbat@apastron.lb.shuttle.de

(URL:  http://www.lb.shuttle.de/apastron/distrCom.htm)