Pages: [1] 2
PatHawks
 
BOINCstats SOFA member
BAM!ID: 63474
Joined: 2008-12-30
Posts: 150
Credits: 492,666
World-rank: 308,096

2009-01-21 21:23:29

To Completion: 255:45:12

WTF kind of joke is this!? For one WU???
That's not serious, is it?
Guest

2009-01-21 21:38:51

To Completion: 255:45:12

WTF kind of joke is this!? For one WU???
That's not serious, is it?


for sure it is.

try orbit (when the put up new ones), those are even longer..
PatHawks
 
BOINCstats SOFA member
BAM!ID: 63474
Joined: 2008-12-30
Posts: 150
Credits: 492,666
World-rank: 308,096

2009-01-21 21:42:10

I'd better get butt loads of credit...
Sid2
 
Forum moderator - BOINCstats SOFA member
BAM!ID: 28578
Joined: 2007-06-13
Posts: 7336
Credits: 593,088,993
World-rank: 3,277

2009-01-21 23:19:04

I'd better get butt loads of credit...


You won't.

Check out this guide.

Everyone has their favorites, but I believe that QMC [expecially when they toss out the long WU's] is hard to beat for real science and good credits.

I really wouldn't recommend marathons to you for a while.



Saenger
 
Tester - Translator
BAM!ID: 5
Joined: 2006-01-10
Posts: 1735
Credits: 226,879,199
World-rank: 6,523

2009-01-21 23:22:52

To Completion: 255:45:12

WTF kind of joke is this!? For one WU???
That's not serious, is it?

It's serious, but unless other projects with real long WUs they send home trickles every few hours, and you get credits for them already.
For my computer the first estimate is mostly a bit too high, perhaps it's just 200h.
Grüße vom Sänger
PatHawks
 
BOINCstats SOFA member
BAM!ID: 63474
Joined: 2008-12-30
Posts: 150
Credits: 492,666
World-rank: 308,096

2009-01-21 23:27:52

It's serious, but unless other projects with real long WUs they send home trickles every few hours, and you get credits for them already.
So it's a cheap way to hog all my resources?
For my computer the first estimate is mostly a bit too high
It went up to 360h before I aborted.

I really wouldn't recommend marathons to you for a while.
I hear that.
I'm still trying to figure all this stuff out.
Sid2
 
Forum moderator - BOINCstats SOFA member
BAM!ID: 28578
Joined: 2007-06-13
Posts: 7336
Credits: 593,088,993
World-rank: 3,277

2009-01-21 23:36:09

It went up to 360h before I aborted.



That was a short one. . . I actually finished one that ran 1800 hours [24/7 for almost 2 months]. . .

. . . I was 200h into one that was 2700h when a HDD crapped.




PatHawks
 
BOINCstats SOFA member
BAM!ID: 63474
Joined: 2008-12-30
Posts: 150
Credits: 492,666
World-rank: 308,096

2009-01-21 23:38:26

I was 200h into one that was 2700h when a HDD crapped.
Ouch!
Saenger
 
Tester - Translator
BAM!ID: 5
Joined: 2006-01-10
Posts: 1735
Credits: 226,879,199
World-rank: 6,523

2009-01-21 23:42:40

It's serious, but unless other projects with real long WUs they send home trickles every few hours, and you get credits for them already.
So it's a cheap way to hog all my resources?

The deadline is about 1 year, so it won't hog your resources. It behaves quite nice even in multi-project environments.
For my computer the first estimate is mostly a bit too high
It went up to 360h before I aborted.

Why abort it?
Grüße vom Sänger
PatHawks
 
BOINCstats SOFA member
BAM!ID: 63474
Joined: 2008-12-30
Posts: 150
Credits: 492,666
World-rank: 308,096

2009-01-21 23:48:16

Why abort it?

I'm still just trying to get a feel for different projects.
Hard to experiment much with week WUs going all the time.

Was that a dumb move?
Guest

2009-01-21 23:52:11

Was that a dumb move?


dumb?

it would be dumb to run things like that on hosts you are not very sure of being real solid.
noderaser
 
BAM!ID: 13859
Joined: 2006-12-03
Posts: 839
Credits: 357,892,079
World-rank: 4,749

2009-01-22 03:55:49

The few CPDN units I've been able to run all the way through have made it. You don't have to wait until the WU is all the way done to get credit; you get credit at each checkpoint. However, it's not terribly helpful if you want to spread your credit around to different projects, which is why I've removed it from my active projects. But, I may bring it back at some point--with multicore projects you can still spread the credit around a little.
Low_Gap
 
BAM!ID: 61291
Joined: 2008-11-09
Posts: 6
Credits: 463,779,913
World-rank: 3,973

2009-01-22 04:56:55

I am running a workunit on one of my older computers that is currently 3,459:13 hours into the job with 1,019:59 hours remaining and a due date of Sept. 7, 2010. It has a resource share of 300 and generally kicks out 311 credits per day. It will periodically send a 15 MB upload in addition to the normal trickle-up. This is the second one of these workunits that I have run. The prior one terminated before completion due to some kind of error.

I doubt that I will take another workunit from this project. I have been having some doubts as to how productive this project really is. I seem to be drifting more to biochemistry/biomedical type projects. This is just personal taste.


PatHawks
 
BOINCstats SOFA member
BAM!ID: 63474
Joined: 2008-12-30
Posts: 150
Credits: 492,666
World-rank: 308,096

2009-01-22 05:33:20

Looking at how few credits it produces, how much trouble it is to run, and how unclear the impact of the project's goals seem to be, I will no longer be running any CPDN WUs.
mo.v
BAM!ID: 25128
Joined: 2007-05-01
Posts: 280
Credits: 0
World-rank: 0

2009-01-23 12:43:16
last modified: 2009-01-23 13:17:28

CPDN members can select the model type they want for their next download. The selection is in the project preferences part of members' CPDN account. The model types are described here.

We recommend that members should look at the project README collections near the top of the independent forum here. Although the models are very long, it's possible to complete almost all of them by making regular backups of the BOINC folder, or for BOINC v.6 the BOINC Data folder. A selection of methods is suggested in the READMEs. So if a model crashes, a backup can be restored and the model completed.

CPDN has a very active research team at Oxford University whose publications can be reached from this page of the project website. They collaborate with researchers at several other universities including Reading, Bristol and Carnegie-Mellon, as well as with researchers at the UK Met Office. Their research goals are very clear indeed.

Research has been into the prediction of future climate, the analysis and improvement of climate models as a research tool, and the attribution of some causes of climate change ie working out what is causing the observed changes. Data and research from CPDN contributed to the most recent IPCC assessment. Since 2007 there has also been a CPDN results website from which climate researchers in any country can download free of charge our crunched model results for their own research.

The research currently being done using HADAM models was described last year in the News threads of the CPDN forums.

The current Geoengineering Experiment which uses HADCM models is described here.

The Mid-Holocene Experiment which uses current HADSM MH models is described here.

The planned Millennium Experiment using a new model type is described here.

sorceress
 
BAM!ID: 12734
Joined: 2006-11-19
Posts: 35
Credits: 5,279,760
World-rank: 82,692

2009-01-23 21:27:10

Pat Hawks, like so many paticipants in distributed computing, is only interested in the 'credits'. They spend $1,000's of dollars to build super computers so that they can see how many 'credits' they can reap per second/hour/day etc. Whatever happened to the 'science'? It's the main reason these projects's existence. These scientists have spent 1,000's of hours in creating programs that use enormous amounts of processing power to aid them in their search for answers to their questions and all many participants seem to think about is 'How many credits can I make?' What a slap in the face that must be to them!

As for me, I am in CPDN for the science. I get a nice bouquette of credits everyday and that is enough for me. Think about the science and about the time spent creating this project and 'crunch 'it for the 'answers!. Just my $0.02

PS... I also like the cool screensaver
zombie67
BAM!ID: 1560
Joined: 2006-06-04
Posts: 606
Credits: 183,391,913,133
World-rank: 43

2009-01-23 22:05:19

As a project, I wouldn't care less if you are crunching for the science or the credits. As long as you were crunching, I would be happy and thankful.
Reno, NV
Team: SETI.USA
PatHawks
 
BOINCstats SOFA member
BAM!ID: 63474
Joined: 2008-12-30
Posts: 150
Credits: 492,666
World-rank: 308,096

2009-01-24 00:54:48

Sorry mo.v
I was wrong about the impact of the projects goals. Next time I'll be sure to RTFM before I flame somebody.

Pat Hawks, like so many paticipants in distributed computing, is only interested in the 'credits'...
...What a slap in the face that must be to them!
Yeah, I really came off sounding like a donkey, huh.

Sorry guys. I was wrong about CPDN.
WUs are freakishly long, but I'm sure it's all worth it in the end.
Rakarin
 
BAM!ID: 1019
Joined: 2006-05-30
Posts: 92
Credits: 0
World-rank: 0

2009-01-24 15:11:53

Sorry guys. I was wrong about CPDN.
WUs are freakishly long, but I'm sure it's all worth it in the end.


Long WU's are not for everyone. If you're a bit OCD like me, then long WU's can bother you. Also, it depends on how old and clunky your equipment is. If you have to restore your CPDN work weekly or monthly because of crashes, then yes, perhaps your rig is not right for the project.

My feeling on credits is that they show that my work was validated and useful in some way. (Even errored out WU's can be useful if it does a good output dump.) I have to pay for electricity to run 4 computers and a PS3, and electricity to run the air conditioning year round. (The power grid here is bad, so I regularly replace batteries in my two large UPS's.) Considering the cost of my assorted hobbies, crunching is actually my most expensive hobby. If I don't get credits for a project, that means I'm getting a high error rate. I'm *paying* $10-$20 a month per PC to crunch. If I'm not getting credits, I check my work history. If I'm crunching bad WU's, I switch projects (and check my PC).

I'm not really concerned with the quantity of points, simply that the count rises at a fairly steady rate. While part of me is thrilled that my points total has gotten so high over the years, there's also a matter of justifying expenses I see the beginning of every month when I pay my bills. (We'll just pretend for now that I pay my bills promptly... )
mo.v
BAM!ID: 25128
Joined: 2007-05-01
Posts: 280
Credits: 0
World-rank: 0

2009-01-24 15:17:00


Hi Pat

No harm done, no problem. I first started distributed computing with CPDN Classic so I just accepted long tasks from the outset. But crunchers accustomed to short tasks who attach to CPDN without finding out in advance what's involved are in for a shock.

Anyone who wants to try a model or two without committing their computer to too much work should deselect the HADCM models in their CPDN project preferences as these are the longest. At the moment I'm running two CPDN beta 200-year models that will take nearly 3000 hours each on a fairly fast machine. But none of the public release models are as long.

As Sorceress said, the graphics are brilliant. I've seen the graphics from a similar Met Office model run at much higher definition on a supercomputer in Tokyo and they're even more amazing. This supercomputer apparently occupies the area of 3 tennis courts so the CPDN research doesn't try to compete with that. We produce large ensembles of models and the research is designed differently.

Rakarin is right about different projects suiting different people and machines.
sorceress
 
BAM!ID: 12734
Joined: 2006-11-19
Posts: 35
Credits: 5,279,760
World-rank: 82,692

2009-01-24 18:17:15


No harm done, no problem. I first started distributed computing with CPDN Classic so I just accepted long tasks from the outset. But crunchers accustomed to short tasks who attach to CPDN without finding out in advance what's involved are in for a shock.


A shock is right! And there are other projects like SETI's Astropulse, Cosmology and Einstein that sometimes can take a while to crunch. I enjoy them all and dont mind the time. Most do pay well when they validate too. Seti, CPDN and Einstein were my first projects to process so I'm somewhat partial to them. I have had computer problems and I ran into an ice planet that have caused me to lose some of processing time so I've never finished a CPDN WU as of yet. But I am getting close to one now. I am anxious to see what a finished one pays. Woohoo

One thing important for anyone trying to decide to join a project, is to spend a little time reading the forums and FAQs. And checking the stats of some of the top users. That will give you an idea on the status for the project, what some of the problems might be, and if its the RIGHT one for you. Many projects dont have work, some are application/OS specific and some are not worth the time to connect. CPDN, SETTI, Einstein and some others alway have work, are well maintained and the scienc is really cool. Choose wisely!
Gundolf Jahn
 
BAM!ID: 16586
Joined: 2007-01-07
Posts: 469
Credits: 447,470
World-rank: 323,769

2009-01-24 22:18:34
last modified: 2009-01-24 22:21:32

off topic @sorceress: you have to remove the blank from your signature and correct the userID:

[img]http://www.boincstats.com/signature/user_1222212.gif[/img]

to get this:
ebahapo
 
BAM!ID: 239
Joined: 2006-05-12
Posts: 662
Credits: 18,928,018
World-rank: 35,235

2009-01-24 23:30:55

Anyone who wants to try a model or two without committing their computer to too much work should deselect the HADCM models in their CPDN project preferences as these are the longest.

Or better yet, select just HADAM3 here, the "shortest" one at 15 days, though gobbling 1.5GB or RAM.

HTH
Rakarin
 
BAM!ID: 1019
Joined: 2006-05-30
Posts: 92
Credits: 0
World-rank: 0

2009-01-25 16:01:33

One thing important for anyone trying to decide to join a project, is to spend a little time reading the forums and FAQs. And checking the stats of some of the top users. That will give you an idea on the status for the project, what some of the problems might be, and if its the RIGHT one for you. Many projects don't have work, some are application/OS specific and some are not worth the time to connect. CPDN, SETI, Einstein and some others always have work, are well maintained and the science is really cool. Choose wisely!


Sorceress... thank you thank you thank you!

Ultimately, yes, that's the best advice. If you run a PPC Mac like me, then yes, you have to see what projects will even run. (Some small ones are Windows or Linux only.) The level of support and development are critical. Check the tone of the users, and the tone of the admins. For that matter, see if anyone is posting that looks like they're actually part of the project. To me, I want to know what the science is, what's being done with it, and who will "own" or work with it. (Also note that BOINC'ing, and Folding and GRIDing, are international. Do *not* fall into the "Ugly American" stereotype of assuming poor grammar shows a lack of intelligence. I've seen kids (mentally, probably physically) do that, and it makes me weep. If communicating to an international crowd, be careful no to "dumb it down", as that can come across, but use short sentences, be concise and focused, stay on topic, and avoid metaphors or exaggerations. Also, the other party may be having someone translate back and forth, so be generous with time when expecting a response.)

Plan to try a couple work units, a sometimes a particular chipset / OS / anti-malware / firewall / Mercury going into the wrong astrological house while retrograde will make things just not work. I crunch for Superlink. It works fine on my Linux box(es, now one), and my G5 Mac, but in 8 months I've only had 3 or 4 work units work on my Windows XP / AMD Athlon PC. The last time I attempted to run it, on that PC, I had a ~50% failure rate, as opposed to 100%. Don't know why, but it runs on other boxes. Anyway, if you're unsure, link one box. If the results are inconsistent, but it sounds like the project is developing, leave just one box attached, and set the time slice to 50 or 25, and check results on the web site weekly or so.

Actually, that's another good point. When you start a project, check your profile on the web site every few days at first, then weekly for a bit. Check your pending credit and results per computer. Make sure your results are getting results.

The amount and rate of work is important if you run multiple projects. If all your projects are intermittent, you'll have a feast or famine scenario.

Well, I'm prattling on here... Enjoy your years of BOINC'ing! As Sorceress said, do some research, and ultimately choose what interests and suits you.

PatHawks
 
BOINCstats SOFA member
BAM!ID: 63474
Joined: 2008-12-30
Posts: 150
Credits: 492,666
World-rank: 308,096

2009-01-25 17:25:46

When you start a project, check your profile on the web site every few days at first, then weekly for a bit. Check your pending credit and results per computer. Make sure your results are getting results.
Couldn't have said it better myself.
Pages: [1] 2

Index :: The Projects :: CPDN
Reason: