> Rock on. I hear you. One must wonder, though, if CPDN's choice to forego
> redundant checking will bite them at some point.

If, and that is a big if, I understand their process it is not actually foregone. The models we process take a long time to go from the start condition to an end condition. Basiclly a full month or more per model. Because of the demand, the rudundant processing is not a smart option because of the current lack of processing power.

This is one of the reasons that many of us are advocating that more people go there and sign up. If we could get all of the SETI@Home people to sign up we might be able to kill two stones with one bird. however, I digress ...

When the model is complete it is phoned home as a small result file so save bandwidth. If the model is interesting, the project team asks the participant to upload the really big results file for additional analysis. Even without that, they could put that model back in the hopper to have someone else run it and get results. So, they have cake and eat it too ...
<p>
For BOINC Documentaion: Click Me!

> It wasn't that long ago that there where only about 9000 Clients over at CPDN,
> now there is close to 20,000 over there according to the Project Stat's. That
> tells me there are a lot of People that simply don't like the Color of the
> Ship over here anyway.


I attached to CPDN, too. But maybe I'm a little bone or I didn't understand something wrong. I downloaded the client and one (!) WU from CPDN and when it started to crunch I was shocked about the time to process. It had been more than 1540 hours. maybe I was paniked or something similar but I detattched from CPDN immediatelly. My account is still alive at CPDN but I really don't know if I attach to it any more. I'm running a 1.2 GHz AMD machine with 384 MB RAM and 100GB HDD with dialup modem. AND I would say: A WU that's need over 10 hours to be crunched is useless time for ME 'cause my PC is running in my bedroom for not longer than 8 hours a day.

Maybe someone who's attached some longer to CPDN can tell me how long the WU of it are processed in the middle.


Best wishes from germany.

Michael
<a> [/url]

> I attached to CPDN, too. But maybe I'm a little bone or I didn't understand
> something wrong. I downloaded the client and one (!) WU from CPDN and when it
> started to crunch I was shocked about the time to process. It had been more
> than 1540 hours. maybe I was paniked or something similar but I detattched
> from CPDN immediatelly. My account is still alive at CPDN but I really don't
> know if I attach to it any more. I'm running a 1.2 GHz AMD machine with 384 MB
> RAM and 100GB HDD with dialup modem. AND I would say: A WU that's need over 10
> hours to be crunched is useless time for ME 'cause my PC is running in my
> bedroom for not longer than 8 hours a day.
>
> Maybe someone who's attached some longer to CPDN can tell me how long the WU
> of it are processed in the middle.

Depending on the speed of your machine that sounds about right. Mine list at 500 hours. If you call up the graphis when it is processing you can see that they are modeling the climate. Kinda like SimEarth ...

So, it is still ok to keep one there for those times when projects run slow refreshes on the work. I have one cp.net work unit on all my computers, 3 active and running projects and I have work being accomplished on all of them ...

I seem to be hitting the LHC@Home maximum number of work units per day, and without predictor there to fill in it is a LITTLE nervous ... but, actually the round robin processing is working pretty well. But BOINC View does not have a summary line on some of the windows where it would be nice (like to know on a per computer basis just how full the queue is based on assumed runtimes).

So, restart your engine and get another model... who knows, maybe this server outage may take longer than we think ...
<p>
For BOINC Documentaion: Click Me!

I attached to CPDN, too. But maybe I'm a little bone or I didn't understand something wrong. I downloaded the client and one (!) WU from CPDN and when it started to crunch I was shocked about the time to process. It had been more than 1540 hours.
=========

Thats about right, My P3 850 showed 1600 hours to competion at the start of the run, my 3.4 EE showed about 550 hours...the other Pc's I have fall in between the 2...

> ... my 3.4 EE showed about 550 hours...

I been contimplating upgrading one of my slower computers with one of those EE processors...

A little green with envy ... :)

Though I like the Macintosh better ....

<p>
For BOINC Documentaion: Click Me!

Don't worry Riker! :) I have one poor machine that is going to take a bit over 2000 hours to finish a CPDN WU.

The CPDN folks know that their model runs are long and they've adjusted the expiration times accordingly. I think you technically have ~11.5 MONTHS from the time you download a CPDN model run 'til you need to get it completed.

Admittedly, it will take your machine (running at 8 hours per day) ~6 months to complete a WU, so make sure you set your preferences to ~50/50 SETI/CPDN, but once you do that, you'll NEVER have to worry about your computer not having anything to do if you can't get a fresh supply of SETI. ;)

> ... my 3.4 EE showed about 550 hours...

I been contimplating upgrading one of my slower computers with one of those EE processors...

A little green with envy ... :)

Though I like the Macintosh better
==========

Mucho Bucks for the EE CPU Paul, whether it's worth it or not who knows, I just wanted one so I went ahead and splurged...

Poorboy,

> Mucho Bucks for the EE CPU Paul, whether it's worth it or not who knows, I
> just wanted one so I went ahead and splurged...

Well, when I looked at the price they were only about $1,000 or so... since I have the rest and all I would need right off would be the processor, MB, and RAM, couild up-engine one of my P4 2.8GHz machines ...

But right now I think I am just going to wait until next year and get another dual G5, but this one will be the 2.5 or better ...

There I already have the 23" monitor so I don't need one of those right off ... the Mac will bump down the food chain and have to get by on a 20.1" NEC LCD 2010x ...

This kind of thing is why I am so interested in the results of the optimization efforts. With SETI being open source we should see some tailored clients and then you should be able to find out what processor, with best optimization does the work fastest. For my work station I will stay mac now I think... But I have those other computers that for the most part only do the BOINCing.
<p>
For BOINC Documentaion: Click Me!

Well, when I looked at the price they were only about $1,000 or so... since I have the rest and all I would need right off would be the processor, MB, and RAM, couild up-engine one of my P4 2.8GHz machines ...
===========

I hear you Paul. I have a P3 850 & a P4 1.9 that are holding up the Mini Farm, I would like to do something with them before the year is out...

Myself I would think the Money would be better spent on a Regular P4 3.4 @ $325 or even the P4 3.6 @ $480, I would think a Regular P4 3.6 would give the P4 3.4 EE a good run for the money and it wouldn't cost as much.

> I hear you Paul. I have a P3 850 & a P4 1.9 that are holding up the Mini
> Farm, I would like to do something with them before the year is out...
>
> Myself I would think the Money would be better spent on a Regular P4 3.4 @
> $325 or even the P4 3.6 @ $480, I would think a Regular P4 3.6 would give the
> P4 3.4 EE a good run for the money and it wouldn't cost as much.

Well, sometimes I just like to go for the gusto as the comercial once said...

Usually, I go and buy a high end set of parts as they will give me a longer life. For example, I have a total of 3 Pentium 4 at 2.8 GHz and they are still respectible systems. I coould let them go on for another couple of years and still be getting good results from them.

It is like the monitors. I had an old NEX Multisync 21" monitor (2-man lift) that lasted me for over a decade ...

I have had one of the 2010 for over 5 years by now. When I got the Mac I got the 23" screen as it is expected to last me for a fair number of years. Grant you, I would have gotten the 30" if it had been available ...
<p>
For BOINC Documentaion: Click Me!

> Don't worry Riker! :) I have one poor machine that is going to take a bit
> over 2000 hours to finish a CPDN WU.
>
> The CPDN folks know that their model runs are long and they've adjusted the
> expiration times accordingly. I think you technically have ~11.5 MONTHS from
> the time you download a CPDN model run 'til you need to get it completed.
>
> Admittedly, it will take your machine (running at 8 hours per day) ~6 months
> to complete a WU, so make sure you set your preferences to ~50/50 SETI/CPDN,
> but once you do that, you'll NEVER have to worry about your computer not
> having anything to do if you can't get a fresh supply of SETI. ;)
>
So thank god I'm not that dumb I believed. I'ld have run predictor, too, but unless they didn't upgrade there client to version 4 I'm unable to get some work. And what's about seti client 4.09? Is it stable or are there any probs so it would better to wait for another upgrade? I think I read something about some probs in another thread.


Best wishes from germany.

Michael
<a> [/url]

> So thank god I'm not that dumb I believed. I'ld have run predictor, too, but
> unless they didn't upgrade there client to version 4 I'm unable to get some
> work. And what's about seti client 4.09? Is it stable or are there any probs
> so it would better to wait for another upgrade? I think I read something about
> some probs in another thread.

Yes, it can be shocking to see your first CPDN workunit, but you do get incremental credit for it as it's progressing along. Roughly every 1.3% of the WU, you'll get a 'trickle' credit on the CPDN site.

Many of us are waiting for Predictor to get going with Version 4, but their original purpose for doing the first predictor project is done (CASP-6, more info on their site about that), so they have the luxury of taking a bit of a pause and they'll likely come back with a slightly changed project based on what they learned from the first run. Hopefully within a few weeks they'll be back going again.

By 4.09, I assume you mean the BOINC client and not the SETI application that runs within it (which is still the same 4.03 that they've been running since going to version 4 of BOINC). In answer to your question, I have it running on 7 different machines, both as a service on some and as a GUI on a couple and I have no problems at all (knock on wood). Once in a while, the clients are not able to get work, but with a 'Contact servers' setting of 2 days, I've never had one completely run out of work since their >week long outage a few weeks ago. All the clients are set to do part SETI/ part CPDN though, so even if I do run out of SETI for a short time, I'll just do 100% CPDN for a while.

I hope that covers everything for you. :)

> work. And what's about seti client 4.09? Is it stable or are there any probs
> so it would better to wait for another upgrade? I think I read something about
> some probs in another thread.

The problems that I have talked about predate 4.05, so ... 4.09 has nothing new, it just came up now ...

As we explore and get more experience we are now showing up problems we would have seen in the past because we were having so many other things going wrong that it "masked" these issues.

To me, that is encouraging because we are starting to see "wierdies" now ....

:)
<p>
For BOINC Documentaion: Click Me!

> > The cp.net project is using a value of 1, and because of the long run
> times of
> > the models gives out small "trickels" of earned credit. The remaining
> > projects currently available are all using some variation of the
> SETI@Home
> > redundant computing process.
>
> Rock on. I hear you. One must wonder, though, if CPDN's choice to forego
> redundant checking will bite them at some point.

It took me some searching, but here's the answer to that:

we send some duplicate workunits for validation of results, but we aren't using the SETI "paradigm" of redundant workunits since models will in an ensemble, even given the same initial conditions, will not perform exactly due to the non-linear chaotic nature of climate systems.

----------------------
Jordâ„¢

I am eagerly looking forward for the next level...

A second stage BOINC Project ... Can you imagine?

A project that, unlike Astropulse, does not just re-hash the collected data, but a project that takes the outputs of several BOINC projects and uses that data as its source information ...

:)
<p>
For BOINC Documentaion: Click Me!