climateprediction.net (CPDN) home page
Thread 'Workunit no.708050'

Thread 'Workunit no.708050'

Message boards : Number crunching : Workunit no.708050
Message board moderation

To post messages, you must log in.

AuthorMessage
old_user94012

Send message
Joined: 23 Aug 05
Posts: 12
Credit: 41,808
RAC: 0
Message 18685 - Posted: 24 Dec 2005, 3:24:18 UTC

It disappeared from my work list in the late October and I was crunching the no.333311 since then. Today I checked my result list on the site and found out the lost wu is still there without anyone else crunching it. Is it ok?


Welcome To Team China!
ID: 18685 · Report as offensive     Reply Quote
Les Bayliss
Volunteer moderator

Send message
Joined: 5 Sep 04
Posts: 7629
Credit: 24,240,330
RAC: 0
Message 18688 - Posted: 24 Dec 2005, 4:12:15 UTC

Possibly. We can\'t check because you have your computers hidden.

ID: 18688 · Report as offensive     Reply Quote
old_user94012

Send message
Joined: 23 Aug 05
Posts: 12
Credit: 41,808
RAC: 0
Message 18696 - Posted: 24 Dec 2005, 10:59:50 UTC - in response to Message 18688.  

Oh, I just reseted the \'Show Your computers\' to yes. You may check it now.
Possibly. We can\'t check because you have your computers hidden.





Welcome To Team China!
ID: 18696 · Report as offensive     Reply Quote
Profileold_user5994

Send message
Joined: 31 Aug 04
Posts: 239
Credit: 2,933,299
RAC: 0
Message 18699 - Posted: 24 Dec 2005, 11:25:21 UTC
Last modified: 24 Dec 2005, 11:25:46 UTC

Well, you can check it for yourself you know ...

But clicking through it looks to me like it was not re-issued and is not in work.

From Your Account, click on Your Computers, then the link for your computer, then down at the bottom the link for results (of which you have 3), click the 3, then the work unit you are interested in ...

And it has not been re-issued. Since they suspended issuing Slab models, no telling when or if it will be issued ...

The last trickle seems to have been on 24 Dec 2005 03:20:35, by you, for the new model ...
ID: 18699 · Report as offensive     Reply Quote
Les Bayliss
Volunteer moderator

Send message
Joined: 5 Sep 04
Posts: 7629
Credit: 24,240,330
RAC: 0
Message 18701 - Posted: 24 Dec 2005, 11:48:29 UTC

And, according to the model that failed, you are using BOINC version 4.45, which has a bug.
Best to upgrade to the latest version.

ID: 18701 · Report as offensive     Reply Quote
old_user94012

Send message
Joined: 23 Aug 05
Posts: 12
Credit: 41,808
RAC: 0
Message 18721 - Posted: 25 Dec 2005, 8:34:54 UTC - in response to Message 18699.  

Paul,

Thanks for your detailed explanation. So for now the project only issues worksunits for the sulphur model?

Well, you can check it for yourself you know ...

But clicking through it looks to me like it was not re-issued and is not in work.

From Your Account, click on Your Computers, then the link for your computer, then down at the bottom the link for results (of which you have 3), click the 3, then the work unit you are interested in ...

And it has not been re-issued. Since they suspended issuing Slab models, no telling when or if it will be issued ...

The last trickle seems to have been on 24 Dec 2005 03:20:35, by you, for the new model ...




Welcome To Team China!
ID: 18721 · Report as offensive     Reply Quote
old_user94012

Send message
Joined: 23 Aug 05
Posts: 12
Credit: 41,808
RAC: 0
Message 18723 - Posted: 25 Dec 2005, 8:42:08 UTC - in response to Message 18701.  

Les,

Thanks for your advice, I\'ll upgrade to the new 5.2.x version. Any detail about the bug with 4.45?

BTW, Happy Christmas to all you guys:)

And, according to the model that failed, you are using BOINC version 4.45, which has a bug.
Best to upgrade to the latest version.





Welcome To Team China!
ID: 18723 · Report as offensive     Reply Quote
Profileold_user5994

Send message
Joined: 31 Aug 04
Posts: 239
Credit: 2,933,299
RAC: 0
Message 18729 - Posted: 25 Dec 2005, 18:11:00 UTC

At the moment, yes. Sulfur is the current expiriment.

I can\'t get my head around what CPDN is doing exactly, not that it is that difficult I suppose, just a mental block (I have the same problem at Rosetta@Home too ...).

Les, or one of the other gurus might be better at this, but, *NY* current expectation is that we will do Sulfur now, then next will be \"coupled\" which will take the outputs we have done and do something else to them.

I like \"slab\" only from the sense that I like to FINISH the models and with run times in the months the risk of a failure is obviously higher. THAT being said, I have been having very good luck with all the models on Windows ... So, I hope to finish my FIRST Sulfur models and my last two slab models in the next two weeks ... Yea!

WIth regard to the version, I think the problem Les was talking about is the one where the application could download more work than it should. I am not sure it is fixed in even the latest versions. IF THAT IS the problem Les is referring to, just set CPDN to \"No New Work\" and a when the model goes up, enable it long enough to download a new model and then turn it off again.
ID: 18729 · Report as offensive     Reply Quote
Les Bayliss
Volunteer moderator

Send message
Joined: 5 Sep 04
Posts: 7629
Credit: 24,240,330
RAC: 0
Message 18732 - Posted: 25 Dec 2005, 21:00:07 UTC

Morning everyone.
There IS the problem with 4.45 downloading excessive numbers of wus, but the bug I had in mind is the benchmark bug.
Every 5 days, BOINC pauses the science app, and runs some benchmark software. With 4.45, there is a timing problem. When BOINC is finished, it waits for the app to restart. But the app has already timed out waiting for BOINC to finish.
So BOINC just sits there waiting. It LOOKS as it it is working, but the science app isn\'t running. And for people who only look at the computer every few days, there is a lot of lost time. The cure, apparently, is to stop BOINC, and then restart it.
Which is why it\'s a good idea to move to the new version.

I was going to talk about slab / suphur here, but it\'s turned into a mini novel, so I\'m putting it in a separate post, with a more obvious title.

ID: 18732 · Report as offensive     Reply Quote
old_user94012

Send message
Joined: 23 Aug 05
Posts: 12
Credit: 41,808
RAC: 0
Message 18736 - Posted: 26 Dec 2005, 4:16:25 UTC

I checked the stdoutdae.txt around the late Oct. It seems that the v4.45\'s benchmark bug stands a good chance of causing the problem with the lost wu.

Thanks again to both of you for clearing my doubts ;)


Welcome To Team China!
ID: 18736 · Report as offensive     Reply Quote
Profileold_user5994

Send message
Joined: 31 Aug 04
Posts: 239
Credit: 2,933,299
RAC: 0
Message 18751 - Posted: 26 Dec 2005, 17:59:09 UTC

Yin Gang,

I do *NOT* recommend getting into the upgrade to the latest and greatest without need.

But, *I* have found that there are \"golden\" versions that have few if any bugs. 4.19 was one, then there were a couple in the 4.2 and 4.3 series. Aside from the restart bug 4.45 was good, but 4.72 was the best of the them all for me until 5.2.13 ...

If you only have one system it is a little harder to know when to make the changes. But if there are no significant problems FOR YOU, then stick with what works for you. If you are not sure, ask. In general though, when you get BOINC working with one version, the upgrade is relatively painless.
ID: 18751 · Report as offensive     Reply Quote
old_user94012

Send message
Joined: 23 Aug 05
Posts: 12
Credit: 41,808
RAC: 0
Message 18793 - Posted: 28 Dec 2005, 8:13:56 UTC
Last modified: 28 Dec 2005, 8:15:51 UTC

Hi Paul,

I\'ve upgrade all my crunching systems from 4.45 to 5.2.13 and the 5.2.13 seems faily good until now:)

Thanks again, no need - no upgrade! :)


Welcome To Team China!
ID: 18793 · Report as offensive     Reply Quote
Profileold_user5994

Send message
Joined: 31 Aug 04
Posts: 239
Credit: 2,933,299
RAC: 0
Message 18808 - Posted: 28 Dec 2005, 16:57:44 UTC

We all love a happy ending... :)

Today I thought I \"lost\", or crashed, a 34+ day old model ... ouch! But, it seems to have restarted ok again ... whew! Happy endings, I love em ...
ID: 18808 · Report as offensive     Reply Quote

Message boards : Number crunching : Workunit no.708050

©2024 cpdn.org