Clustering ability???

[RogerAdam]

I have some modeling apps that can run in a clustered environment, from the way I interpret this when explained, the CPU's all work together on 1 instance of an application -plz don't hurt my brain if I'm wrong, if I interpreted this correctly, will any future client have such an ability? -Logistics would be infinately easier, and I imagine duplication of work under these conditions would be very limited if not elimanated.

[Aegion]

Originally posted by RogerAdam
I have some modeling apps that can run in a clustered environment, from the way I interpret this when explained, the CPU's all work together on 1 instance of an application -plz don't hurt my brain if I'm wrong, if I interpreted this correctly, will any future client have such an ability? -Logistics would be infinately easier, and I imagine duplication of work under these conditions would be very limited if not elimanated.
You will need to run multiple instances of each client for each cpu. deling apps benefit from clustering since they can render a specific scene far more rapidly that way, saving the person running the program such a long wait, individual structures are already processed extremely quickly. The benefits from clustering simply are not worth it with this sort of application. I'm not sure the number of duplicated work would be reduced under the circumstances you mention. I believe most of the duplicate work comes from entirely different sources, not from the program being run on different computers in a farm.

[RogerAdam]

Thanks for your answer. I'm not sure if I'll DC anymore, to much semantics involved (on a personal level), don't like the unreceptive part coming from people.

[MAD-ness]

Roger: while they are your computers and it is your choice, I strongly suggest you re-think this.

Obviously you will have a ton of work invested in running DC (moreso than any of the rest of us due to the scale involved), but I think that it is worthwhile.

I think that the reasons you are considering not running DC are the wrong reasons to not run it.

I have NEVER found an online community that didn't squabble and argue. The true test of an online community is how they respond afterwards.

One problem with forums and other similar community places is that both the qualified and unqualified have equal voices. The petty can speak as loudly as the virtuous.

There is definitely a high signal to noise ratio, but that is something which you can learn to filter automatically (most of us have already done so). It is really a case of throwing out the baby with the bathwater.

If you feel that the effort involved in getting the DC clients running isn't worth it, then scale down and only run on as many computers as you can manage with relative ease.

Or, get them up and running, create your own team, and then never read the forums unless you need tech help.

Many people do this. They find a project they consider worthwhile, then they leave it to crunch away and avoid the whole community scene and the competitiveness.

I think I speak for most of us here when I say that we value your potential contribution to the project more than we value any potential contribution to the STATS.

In fact, a few at Ars are already nervous about how we would possibly keep first place if you brought most of your machines to DF...but the consensus amongst us is that it would be a "good" problem to have.

This project IS worthwhile. :)

[KWSN_robegeor]

MAD-ness speaks the truth. :thumbs:

I also recommend you give DF a try. And don't let some jerks (a very small minority in this project) sway your opinion. Let your machines do the talking if nothing else, but whether a person is a huge producer or not, this project has been the best run one that I have tried yet.

Sure, the team jumping can be a pain (if one is prone to become upset by such trivial matters), but, as a person who spent much time with Seti and G@H (Hey, Stefan, :moon: ), I have to say I really like this project and would love to see more horsepower (frogpower?) added to it. Regardless of team affiliation.



-Robert

Ni!

[pointwood]

I can only say the same - please don't stop before you even started :(

Originally posted by MAD-ness In fact, a few at Ars are already nervous about how we would possibly keep first place if you brought most of your machines to DF...but the consensus amongst us is that it would be a "good" problem to have. [/B] The competition is exactly what drives a lot of people. It's the competition that makes it exciting. Right now, thanks to all our great members, we are beating everyone else in this project. I would love to see someone challence us in daily production!

Bring it on Roger or anyone else for that matter! Catch us if you can :moon: :moon: :moon: :D :thumbs:

[shifted]

I've participated in many DC projects (this is my fourth), but i do it mainly because i hate seeing cycles go wasted... Granted, there is a bit of competition for personal stats, but i don't go out of my way. I've never joined a team, because i feel the whole team thing is a waste of time (but that is just my own opinion).

[runestar]

Originally posted by shifted
I've participated in many DC projects (this is my fourth), but i do it mainly because i hate seeing cycles go wasted... Granted, there is a bit of competition for personal stats, but i don't go out of my way. I've never joined a team, because i feel the whole team thing is a waste of time (but that is just my own opinion).

Well, the really good thing about is a team is (assuming its fairly active) you get a place where you can bounce off ideas off other people as well as getting a more immediate response to problems.

For some the stats is a secondary thing, but its nice being in a team, seeing familar faces, so to speak. Also, often times they have a collection of resources and/or links to resources related to distributed computing projects rather than you have to go out on the web and search for things. On that, you can also see what tools the team likes and doesn't like, and why.

If you're avoiding D.C. projects because of the competition... well, I hate to say it... you are going to be avoiding them for a long time... while the greater good is certainly noble in itself, it is the competition that helps fuel interest and participation in these projects. So the science isn't the main reason for a lot of people... it doesn't make the project any less valid or useful... actually quite the opposite. You get more people to contributed more results, and more results equals faster comparasions and new things to explore.

If you don't want to get involved in the competition, don't look at the overall stats. =) Although from a non-competitor point of view its still certainly quite interesting to see the computing power poured into the project.

Here's what I see as good about this project:

1. Good for a variety of machines. Even though officially a PII is recommended, this project works well with Pentiums on the PC side.

2. Good rapor with the founders, particularly Howard. Reasonably open to suggestions and requests... If an idea has merit, he will at least consider it... if it doesn't, he doesn't waste time not telling you so. And he's not snotty about it... sometimes there are some reasonable suggestions but implementing would be little or no enhancement to the running of the client or the project.

3. A reasonably stable client. No beta testing in the middle of using a finalized client for instance.

4. A fairly secure system against hacks and cheats. Although certainly not encouraged, you can try out for yourself and see its rather difficult to do so. Unlike SETI where the project has turned a completely blind eye to the cheating and hacking despite notices to it. Makes you wonder how good the science really is if people are rampantly cheating.

5. Non-commericial project. While I certainly applaud the efforts of commercial companies to utilize D.C. ... there is the spirit of running non-profit projects.


TTFN,

RuneStar½

[shifted]

Yeah, i'm finding this community here at free-dc/DF boards to be nice :) It *is* definitely helpful having a forum.

I tend to avoid overall stats because i don't have the resources to throw at things and "win" ;)

Myself, i absolutely will not run a commercial project, unless i'm getting paid for it. Helping out non-profit is totally cool though.

[runestar]

Unless you run the project for a couple weeks, the stats are meaningless to you... I mean you goto DF and you come from SETI... you think everybody here has clusters... then you start moving up and its not so far-fetched having these big numbers on your own. =)

RS½