Distributed Computing

[R_Head]

Posted this on the Mandriva Forum, nobody have not replied yet. I hope someone has some input on this.

Were I work, we have a lot of stress analysis/calculations to make. For a single computer could take days to accomplish such task.

I am contemplating the idea of Distributed Computing in order to utilize all the PCs on the plant to do number crunching while not in use or non peak hours.

In the past, I used such programs to fold proteins for overclocking competitions, so, anybody knows a software that can do the same thing, that is Open Source and can work for MS and Linux systems? What I am trying to pitch is for each node to do some Stress Calculations to increase the turn around time.

[viking60]

Uh you caught me off guard there. Anyone else? (I do not understand rocket science ) We might even have to go all the way up to Guru level...

[R_Head]

BOINC comes to mind but I do not know if any other one out there that is better. Mathematica is closed source, but I have no experience except for BOINC as far folding on overclocking test/competitions.

[dedanna1029]

Are you looking for something that would do like the World Community Grid, on an industrial plant scale?

[R_Head]

For what I read, Grid type is not suited for us but more of a local peer to peer distributed computing.

I might pitch BOINC and Mathematica if they want to pay for it.

[Snorkasaurus]

I know I am necromancing an[other] old thread, but the idea of distributed computing sounds really interesting. I have heard of Beowulf systems but unfortunately have never had the time to dive all the way in to the topic. Has anyone around here ever attempted such a project? I'd love to hear about it!



Snork.

[dedanna1029]

For what I read, Grid type is not suited for us but more of a local peer to peer distributed computing.

Wow, sorry P., I'd forgotten about this thread altogether until I saw the post from Snork here.

In fact, World Community Grid is indeed a distributive computing project, similar to BOINC. What exactly it uses insofar as its own programming goes, I'm not sure. Wish I did, because how it's implemented is really super, IMO.

SETI@Home is another one btw.

There is a site dedicated to active distributive computing, btw. You might want to have a look. There is also a list of projects at Wikipedia.

There's a bit more info about what these projects are using here, but they don't name the software specifically. I'm still trying to get down to that.

Yes, necrobumping again, but this is a good subject, IMO, and should be kept alive. It's one of the best applications of computing there is.

[dedanna1029]

I also think it would be really cool if you could take something like this and use it for non-medical purposes; adapt it to suit yours. So far, that's all I'm seeing distributed computing used for is for medical research, which of course is for a very noble cause, but I also think it would be cool to use distributed computing for other purposes as well.

[Snorkasaurus]

SETI@Home is another one btw.

I ran the SETI@Home client for a long time, I gave up when they switched to BOINC. The BOINC client is more of a resource hog than the original SETI client and the SETI user stats became much harder to sort through once they switched. Unfortunately it looks like a lot of projects are using BOINC (or just disappearing).

I also think it would be really cool if you could take something like this and use it for non-medical purposes; adapt it to suit yours. So far, that's all I'm seeing distributed computing used for is for medical research, which of course is for a very noble cause, but I also think it would be cool to use distributed computing for other purposes as well.

I was hoping that someone here may have actually setup their own (non-BOINC) distributed computing project and had setup a number of machines to share the load. I have seen some references to using rpi's for distributed computing, but I can't imagine it would be worthwhile unless it specifically used the GPU.

s.

[dedanna1029]

I know what you mean Snork. I ran these projects (all of them at various times), and it ran fine for me at the time. With BOINC, I just set it to use only X portion of my idle computing power, and that sufficed. At times it would get to be a bit much, but it was handleable. I would also set it (sometimes) to only run at night when the computer wasn't being used (this at alternating times of using these things). I don't know if these options are still available with it but I'm about to find out because I want to resume a few of the projects I was contributing to.

Edit for addendums: Oh, and on non-BOINC distributed computing projects, I'm finding nothing. I'm assuming R_Head had run searches and found nothing too. I've been searching for a good couple hours but have come up with very little. There are only three main places that even have distributed computing software, and that seems to be where the projects stem from such as BOINC.

I'd love to see someone take a stab at it though.

P.:

This work was supported by the Air Force Office of Scientific Research under grant AFOSR 85-0252.

You may be interested in this. (.pdf)

[dedanna1029]

P., there's also this.

[dedanna1029]

*Bump for viking60*
What do you think?