climateprediction.net home page
I\'ve had it

I\'ve had it

Questions and Answers : Windows : I\'ve had it
Message board moderation

To post messages, you must log in.

AuthorMessage
old_user1154

Send message
Joined: 25 Aug 04
Posts: 6
Credit: 205,686
RAC: 0
Message 6958 - Posted: 15 Dec 2004, 6:13:33 UTC

With 500 hrs done and less then 100 to go, CPDN decided to quit on me again.
It looks that I'll never get any WU's done.

I've detached CPDN on my main PC.
I still have 1 PC running, and will let it go to the end (probably not), and then I'll stop that one 2.

If someone has news of a more stable version of the CPDN client, give me a shout.

Oh, and don't give me the 'your system is unstable' lecture, because both systems are top of the line, standard-running, virus- and malware- free systems.

Am I angry?
No, just dissapointed that i waisted so many hours of CPU-time without any contribution to science.

Good luck to the rest of you.
ID: 6958 · Report as offensive     Reply Quote
keputnam

Send message
Joined: 31 Aug 04
Posts: 26
Credit: 3,885,681
RAC: 1,651
Message 6972 - Posted: 15 Dec 2004, 17:43:30 UTC

I am almost to the same point

All other projects (Seti and Pirates for now, LHC and Predictor before they shut down) are rock solid.

I don't think that I've ever gotten farther than about 40% when CPDN throws a Fortram Error or a program exception or for some reason trashes the current WU and downloads another.

The only thing that CPDN really has going for it is that right now, it and Seti are the only ones with work to do. There for a while, they were the only project that actually was giving out work

ID: 6972 · Report as offensive     Reply Quote
old_user23880
Volunteer tester

Send message
Joined: 10 Oct 04
Posts: 223
Credit: 4,664
RAC: 0
Message 6991 - Posted: 16 Dec 2004, 23:42:51 UTC

Hi Sting and Ken

Have a look at what I wrote in answer to Steve in his Exit Code -5 thread further down this same message board. What Tolu says there is v. significant.

I'm now back on classic cpdn, which is working perfectly just as it did before I tried boinc. I will stick with this until boinc becomes more Athlon-compatible or until I need a new processor. What I don't know is whether you can run boinc (for Seti etc) and cpdn classic simultaneously.

The matter has been extensively discussed on the classic forum. See the 1 in 8 thread. Or, for my experiences in allegorical style, the Server in Switzerland still Down thread.
__________________________________________________

ID: 6991 · Report as offensive     Reply Quote
Profile Andrew Hingston
Volunteer moderator

Send message
Joined: 17 Aug 04
Posts: 753
Credit: 9,804,700
RAC: 0
Message 6992 - Posted: 16 Dec 2004, 23:49:58 UTC - in response to Message 6991.  
Last modified: 16 Dec 2004, 23:50:17 UTC

> I will stick with this until boinc becomes more
> Athlon-compatible or until I need a new processor.

It's mysterious though, because I've had no trouble at all running either classic or BOINC flavours with an Athlon CPU, Nor, it seems, have most others, though I realise it's no consolation to those who can't.
ID: 6992 · Report as offensive     Reply Quote
Profile Thyme Lawn
Volunteer moderator

Send message
Joined: 5 Aug 04
Posts: 1283
Credit: 15,824,334
RAC: 0
Message 6995 - Posted: 17 Dec 2004, 8:49:19 UTC - in response to Message 6991.  

> What I don't know is whether you can run boinc (for Seti etc) and cpdn classic simultaneously.

There are a couple of ways it could be done (the first definitely works, the second might).

1) On multi-processor systems (including P4 HT) change the general preferences to restrict the number of processors that BOINC uses. You need to restart BOINC after updating preferences before this takes effect.

2) When I tested CPDN classic and BOINC on the same system the classic version stayed idle if it started up after BOINC. After suspending and resuming BOINC the processing power was shared equally. Testing was on a P4 HT and if might be different on an Athlon.
"The ultimate test of a moral society is the kind of world that it leaves to its children." - Dietrich Bonhoeffer
ID: 6995 · Report as offensive     Reply Quote
keputnam

Send message
Joined: 31 Aug 04
Posts: 26
Credit: 3,885,681
RAC: 1,651
Message 6997 - Posted: 17 Dec 2004, 14:50:32 UTC - in response to Message 6991.  

> Hi Sting and Ken
>
> Have a look at what I wrote in answer to Steve in his Exit Code -5 thread
> further down this same message board. What Tolu says there is v.
> significant.
>
> I'm now back on classic cpdn, which is working perfectly just as it did before
> I tried boinc. I will stick with this until boinc becomes more
> Athlon-compatible or until I need a new processor. What I don't know is
> whether you can run boinc (for Seti etc) and cpdn classic simultaneously.
>
> The matter has been extensively discussed on the classic forum. See the 1 in
> 8 thread. Or, for my experiences in allegorical style, the Server in
> Switzerland still Down thread.
>

But all my machines are Intel based PIII/P4/P4M, so at least in my case, it's not related to Boinc on an Athlon

ID: 6997 · Report as offensive     Reply Quote
old_user23880
Volunteer tester

Send message
Joined: 10 Oct 04
Posts: 223
Credit: 4,664
RAC: 0
Message 7007 - Posted: 18 Dec 2004, 2:16:41 UTC

Hi everybody

Not all Athlons seem to be boinc-incompatible, and I don't pretend to have the slightest idea why. However, my machine not only rejected every single boinc model (even after I'd applied all the advice I could find on the boards, read the entire boinc-cpdn manual, and upgraded to 4.13), but also produced impossible climates in every model even in 1811-12.

There has been discussion elsewhere about how 2 machines running the same model with the same parameters can occasionally produce divergent results. Those results, however, only diverge in phase 2 or 3, which means that some small difference is gradually compounded. My machine produced rubbish from the outset. The reassuring thing is that the calculation errors were detected and the models automatically truncated.

Why models don't sit stably on some Intel systems probably has a different cause. I can only recommend consulting the aforementioned manual, as it does explain lots of points that are not immediately obvious.

Meanwhile, classic continues to perform flawlessly...............


__________________________________________________

ID: 7007 · Report as offensive     Reply Quote
old_user1154

Send message
Joined: 25 Aug 04
Posts: 6
Credit: 205,686
RAC: 0
Message 7011 - Posted: 18 Dec 2004, 12:53:06 UTC - in response to Message 7007.  

So what you are saying is while the classic CPDN is working correctly (never used that one), the BOINC version is flawed?
ID: 7011 · Report as offensive     Reply Quote
Profile Andrew Hingston
Volunteer moderator

Send message
Joined: 17 Aug 04
Posts: 753
Credit: 9,804,700
RAC: 0
Message 7015 - Posted: 18 Dec 2004, 15:37:16 UTC - in response to Message 7011.  

> So what you are saying is while the classic CPDN is working correctly (never
> used that one), the BOINC version is flawed?
>

Well, actually many people did have problems running the classic model. There's nothing to suggest that BOINC is any worse, but it is the most demanding task that most people will ever put their machines through.
ID: 7015 · Report as offensive     Reply Quote
old_user23880
Volunteer tester

Send message
Joined: 10 Oct 04
Posts: 223
Credit: 4,664
RAC: 0
Message 7018 - Posted: 18 Dec 2004, 23:40:05 UTC

Hi Sting

No, I'm not saying that the boinc version is flawed. I agree with Andrew that cpdn is a particularly demanding task 1. because the model itself is complex and 2. the model has to stay put without crashing out for so long (6 weeks+ on my machine working 24/7, though other people with newer, more powerful machines crunch much faster). Lots of people are crunching the boinc version effortlessly.

What I am saying is that I spent quite a lot of time learning exit the model and turn off the machine without losing it, then thinking that my continuing problems were due to version 4.04, and finally concluding when I still had problems with 4.13 that my Athlon and boinc don't like each other and require extended therapy. Other people who have come into cpdn-boinc at a later stage can maybe learn from this experience, not spend so much time losing models, and take quicker remedial action.

They will sort out this problem of models crashing, but I've no doubt it will take time. I'm sure that many problems don't surface until after projects are launched. Beta-testing isn't enough, because the people who sign up are experts, not typical of users like me.


__________________________________________________

ID: 7018 · Report as offensive     Reply Quote
old_user1154

Send message
Joined: 25 Aug 04
Posts: 6
Credit: 205,686
RAC: 0
Message 7020 - Posted: 19 Dec 2004, 21:32:37 UTC - in response to Message 7018.  

> They will sort out this problem of models crashing, but I've no doubt it will
> take time.
>
No problem.
I'm ready to try again when they've sorted out the problems.

I'll check in from time to time.

ID: 7020 · Report as offensive     Reply Quote
old_user2500

Send message
Joined: 28 Aug 04
Posts: 65
Credit: 9,636,280
RAC: 0
Message 7066 - Posted: 26 Dec 2004, 3:50:57 UTC

Most of the crashes on my machines have been associated with heavy video / CPU usage by another program, usually a video game. I have had one crash when Norton AV was running.I have had no crashes under moderate usage situations such as browsing, emailing, or non video intensive games. Civ III is fine. battlefield 1942, Warcraft III, or Zoo Tycoon are not.

I have changed my Boinc preferences to Suspend CPDN when the user is active. I am still keeping BOINC in memory.

I have no idea whether or nor this will result in an improvement in my success ratio but am going to give it a fair run.






ID: 7066 · Report as offensive     Reply Quote
ChrisD

Send message
Joined: 8 Aug 04
Posts: 69
Credit: 1,561,341
RAC: 0
Message 7068 - Posted: 26 Dec 2004, 9:49:27 UTC - in response to Message 7018.  
Last modified: 26 Dec 2004, 9:52:12 UTC

In reply to the following:

> Hi Sting
>
> No, I'm not saying that the boinc version is flawed. I agree with Andrew that
> cpdn is a particularly demanding task 1. because the model itself is complex
> and 2. the model has to stay put without crashing out for so long (6 weeks+ on
> my machine working 24/7, though other people with newer, more powerful
> machines crunch much faster). Lots of people are crunching the boinc version
> effortlessly.
>
> What I am saying is that I spent quite a lot of time learning exit the model
> and turn off the machine without losing it, then thinking that my continuing
> problems were due to version 4.04, and finally concluding when I still had
> problems with 4.13 that my Athlon and boinc don't like each other and require
> extended therapy. Other people who have come into cpdn-boinc at a later stage
> can maybe learn from this experience, not spend so much time losing models,
> and take quicker remedial action.
>
> They will sort out this problem of models crashing, but I've no doubt it will
> take time. I'm sure that many problems don't surface until after projects are
> launched. Beta-testing isn't enough, because the people who sign up are
> experts, not typical of users like me.
>

did You have a look at Your log-files. Do You just happen to have the same problem as me. Take a look at my post.

http://climateapps2.oucs.ox.ac.uk/cpdnboinc/forum_thread.php?id=1472

ChrisD

ID: 7068 · Report as offensive     Reply Quote
ChrisD

Send message
Joined: 8 Aug 04
Posts: 69
Credit: 1,561,341
RAC: 0
Message 7071 - Posted: 26 Dec 2004, 13:11:07 UTC - in response to Message 6991.  

> I'm now back on classic cpdn, which is working perfectly just as it did before
> I tried boinc. I will stick with this until boinc becomes more
> Athlon-compatible or until I need a new processor. What I don't know is
> whether you can run boinc (for Seti etc) and cpdn classic simultaneously.
>
You can run CPDN-Classic and BOINC simultaneously. They usually share the resources nicely 50-50.

ChrisD

ID: 7071 · Report as offensive     Reply Quote
old_user169

Send message
Joined: 5 Aug 04
Posts: 39
Credit: 87,633
RAC: 0
Message 7135 - Posted: 1 Jan 2005, 21:22:33 UTC - in response to Message 7071.  

> You can run CPDN-Classic and BOINC simultaneously. They usually share the
> resources nicely 50-50.
>
> ChrisD


This would bear the risk of a -5 error :-/
ID: 7135 · Report as offensive     Reply Quote
old_user1154

Send message
Joined: 25 Aug 04
Posts: 6
Credit: 205,686
RAC: 0
Message 8232 - Posted: 31 Jan 2005, 11:13:57 UTC - in response to Message 7135.  
Last modified: 31 Jan 2005, 11:19:14 UTC

last WU is nearing competion (only 50 hrs to go)
I keep getting
climateprediction.net - 2005-01-31 08:47:50 - Result 3cl9_100177857_0 exited with zero status but no 'finished' file
climateprediction.net - 2005-01-31 08:47:50 - If this happens repeatedly you may need to reset the project.
climateprediction.net - 2005-01-31 08:47:50 - Restarting result 3cl9_100177857_0 using hadsm3 version 4.04

What will happen if i choose to reset the project?
Will I lose the WU again?
Can I keep going without resetting?
[edit] BOINC 4.13, and no I'm not upgrading now ;)[/edit]

ID: 8232 · Report as offensive     Reply Quote
crandles
Volunteer moderator

Send message
Joined: 16 Oct 04
Posts: 692
Credit: 277,679
RAC: 0
Message 8241 - Posted: 31 Jan 2005, 12:34:00 UTC

Just ignor that message and keep going without resetting.
Visit BOINC WIKI for help

And join BOINC Synergy for all the news in one place.
ID: 8241 · Report as offensive     Reply Quote

Questions and Answers : Windows : I\'ve had it

©2024 climateprediction.net