climateprediction.net home page
Unable to complete WU by deadline: should I continue to participate?

Unable to complete WU by deadline: should I continue to participate?

Message boards : Number crunching : Unable to complete WU by deadline: should I continue to participate?
Message board moderation

To post messages, you must log in.

AuthorMessage
old_user29078

Send message
Joined: 9 Nov 04
Posts: 11
Credit: 93,753
RAC: 0
Message 27833 - Posted: 12 Apr 2007, 16:22:30 UTC

My computer is in use for perhaps 2 or 3 hours a day, and I\'m not willing to leave it on when I\'m not actually actively using it (for cost, environmental and safety reasons).

However, I\'m now processing a work unit which is very large and is going to take a very long time to complete (application: hadcm3 5.15, name: hadcm3pbb_b8on_....). I\'ve done the maths and there\'s simply no way I will be able to complete the work unit by the deadline: in fact, I think I\'ll be lucky to get to 25% done.

I now have to ask: is there any point in my continuing with this WU, knowing that it will not complete? Would the partial data actually be of any use?

I don\'t know the direction in which CPDN is going: if I cancel this WU, will I simply get another one of similar magnitude in its place, or are there smaller tasks available now (or in the near future) which it would actually be worthwhile my attempting?

If my computer\'s limited contribution is not really going to be of any practical use, then I feel that (although CPDN is the BOINC project which most interests me) I would be better off giving a larger (or total) resource share to projects where I can actually complete the work in the required time.

ID: 27833 · Report as offensive     Reply Quote
Profile old_user206071
Avatar

Send message
Joined: 29 Oct 06
Posts: 14
Credit: 99,628
RAC: 0
Message 27834 - Posted: 12 Apr 2007, 17:01:54 UTC
Last modified: 12 Apr 2007, 17:24:18 UTC

I\'m sure a more experienced user will answer your questions too, but with the little knowledge I have about CPDN, I can tell you that you shouldnt take the deadline too seriously, AFAIK its quite irrelevant. Only thing is that your pc needs to send a trickle message within a certain time frame (4 or 6 weeks) otherwise it will be reissued to another host.

About canceling the wu: IMO, \"No!\" - Perseverance is definitely needed for those big CPDN wu\'s!

On the other hand, I took a look at your previous results: They all broke off with a client error. So its likely that this wu will crash/break off too!

What I have also seen: Your other project is WCG! Its my favorite project also! (Although I neclected it a bit because of the PoTM our Team (BOINC Synergy) has... But I\'m always trying to convince my teammates to crunch more WCG! ;) ) If this CPDN fails too, I would suggest donating the precious cpu cycles to a different but also important/beneficial project (WCG defintely is such a project, IMHO!)

Best regards, Shai aka Steffen



PS: Looking for a team? You could support us with WCG... :)


Btw: I also started BOINCing with an AMD Sempron 2800+... Ahh, those were the days... :D Now it\'s a X2 4600+ EE, tried to make it as energyefficient as possible!


My NEW BOINC-Site

Why people joined BOINC Synergy...
ID: 27834 · Report as offensive     Reply Quote
Profile Strathpeffer
Avatar

Send message
Joined: 9 Jan 07
Posts: 497
Credit: 342,899
RAC: 0
Message 27835 - Posted: 12 Apr 2007, 17:16:16 UTC

Hi David, the project admins frequently tell us that all data is useful, but that we should try to go all the way to 2080 if poss. And Shai Halud is quite right in advising you to ignore the deadline - it\'s meaningless where CPDN\'s concerned, just take as long as it takes! It\'s a good idea to back up the folder from time to time though, just in case of a crash - many of us have successfully restored crashed models from backup and gone on to complete them.

Good luck!

You’re more than welcome to
park your BBC credits with team Scotland
and/or join team Scotland in CPDN!
Visit the Scotland team
ID: 27835 · Report as offensive     Reply Quote
Profile MikeMarsUK
Volunteer moderator
Avatar

Send message
Joined: 13 Jan 06
Posts: 1498
Credit: 15,613,038
RAC: 0
Message 27839 - Posted: 12 Apr 2007, 17:23:30 UTC


Even if the model never reaches the end, it\'s still doing useful work. The climate data is uploaded to the servers every model year (in the \'trickle\').


I'm a volunteer and my views are my own.
News and Announcements and FAQ
ID: 27839 · Report as offensive     Reply Quote
Profile astroWX
Volunteer moderator

Send message
Joined: 5 Aug 04
Posts: 1496
Credit: 95,522,203
RAC: 0
Message 27841 - Posted: 12 Apr 2007, 17:38:13 UTC

All three failed Runs were \'Aborted by user\'.

(In boinc Manager, Projects, click on No new work.)

You could be spending a good chunk of your time rerunning work done last time the machine was active. To minimize that, in the Manager, Tasks, Show Graphics, click on the Model, click Z, then 8, to see the countdown counter. When it resets from zero to 432, boinc can be suspended and exited. Restart would then be almost exactly where you left off; otherwise as many as 431 Time steps could be reprocessed [22+ minutes on your machine].)

If running C/L, shut down immediately after a new line entry.

To maximize Model processing in the short time your machine is active, set your Preferences, in \'Your account\', to Yes for leave in memory when suspended and do work when machine is in use. (It will then use whatever you don\'t.)

"We have met the enemy and he is us." -- Pogo
Greetings from coastal Washington state, the scenic US Pacific Northwest.
ID: 27841 · Report as offensive     Reply Quote
old_user29078

Send message
Joined: 9 Nov 04
Posts: 11
Credit: 93,753
RAC: 0
Message 27843 - Posted: 12 Apr 2007, 17:56:44 UTC - in response to Message 27834.  

I\'m sure a more experienced user will answer your questions too, but with the little knowledge I have about CPDN, I can tell you that you shouldnt take the deadline too seriously, AFAIK its quite irrelevant.


OK, it\'s good to know that the deadline isn\'t really a real cut-off deadline.

Although I do wonder, if my computer is only modelling the earlier years in a reasonable timescale, is that data really all that much use? By the time my computer gets around to modelling the \"future\" it really could be a couple of real-world years off, and CPDN will presumably have had plenty of data from faster computers by then?


On the other hand, I took a look at your previous results: They all broke off with a client error. So its likely that this wu will crash/break off too!


Well, I bought a new computer last year, as the old one did have an annoying habit of locking up very frequently, which I really don\'t think helped, as the results show! More recently, it took me several attempts to get my BOINC installation set up the way I wanted it (using BOINC from my OS package repository rather than having to install and configure manually), and I may have inadvertently downloaded and abandoned a few models while tinkering..



Btw: I also started BOINCing with an AMD Sempron 2800+... Ahh, those were the days... :D Now it\'s a X2 4600+ EE, tried to make it as energyefficient as possible!


Oi, that\'s my *new* computer you\'re mocking! ;-/
It may only be last-year\'s entry-level standard, but I can\'t afford a more powerful computer and it\'s about double the clock speed of my old computer (however that equates in real extra performance..). I know I can\'t contribute much in the way of resources, but in theory, every little helps..


ID: 27843 · Report as offensive     Reply Quote
old_user29078

Send message
Joined: 9 Nov 04
Posts: 11
Credit: 93,753
RAC: 0
Message 27845 - Posted: 12 Apr 2007, 18:09:24 UTC - in response to Message 27841.  


You could be spending a good chunk of your time rerunning work done last time the machine was active. To minimize that, in the Manager, Tasks, Show Graphics, click on the Model, click Z, then 8, to see the countdown counter. When it resets from zero to 432, boinc can be suspended and exited. Restart would then be almost exactly where you left off; otherwise as many as 431 Time steps could be reprocessed [22+ minutes on your machine].)


Ah, that would be quite bad :-(

\"Show Graphics\" doesn\'t work for me. I don\'t know if this is a feature missing in Linux or because I run BOINC under its own user and it doesn\'t have authority to open windows on my user\'s X display.


If running C/L, shut down immediately after a new line entry.


Sorry, I don\'t understand what you mean here. \"C/L\" ?



To maximize Model processing in the short time your machine is active, set your Preferences, in \'Your account\', to Yes for leave in memory when suspended and do work when machine is in use. (It will then use whatever you don\'t.)


Yes, I\'ve done that, thanks.

ID: 27845 · Report as offensive     Reply Quote
Profile Strathpeffer
Avatar

Send message
Joined: 9 Jan 07
Posts: 497
Credit: 342,899
RAC: 0
Message 27846 - Posted: 12 Apr 2007, 18:09:44 UTC

David M wrote:
Oi, that\'s my *new* computer you\'re mocking! ;-/
It may only be last-year\'s entry-level standard, but I can\'t afford a more powerful computer and it\'s about double the clock speed of my old computer (however that equates in real extra performance..). I know I can\'t contribute much in the way of resources, but in theory, every little helps..

Indeed and in fact, as well as in theory, every little helps, David - and there are plenty of people with older computers than yours running this experiment, especially the BBC version! Sometimes the computer whizz-kids are inclined to forget that. Do back up your model from time to time though - it\'ll undoubtedly be more satisfying for you as well as more useful to the project if you can reach 2080.

Best of luck!
Visit the Scotland team
ID: 27846 · Report as offensive     Reply Quote
Profile old_user206071
Avatar

Send message
Joined: 29 Oct 06
Posts: 14
Credit: 99,628
RAC: 0
Message 27847 - Posted: 12 Apr 2007, 18:13:49 UTC - in response to Message 27843.  


Oi, that\'s my *new* computer you\'re mocking! ;-/
It may only be last-year\'s entry-level standard, but I can\'t afford a more powerful computer and it\'s about double the clock speed of my old computer (however that equates in real extra performance..). I know I can\'t contribute much in the way of resources, but in theory, every little helps..



[Humour-Mode][Being offended-Mode]Hey I\'m not mocking![/Being offended-Mode][/Humour-Mode]

I loved my Sempron, I was proud as one can be. After 6 years with an AMD K6-2 450MHz, this machine was fast as lightning!

18 months ago, I got my first real full-time unlimited job, so last december I was able to afford this dual-core. It was also with the help of Ebay, cause compared to new CPUs the Socket A Semprons sold for quite some money, especially the 2800+ and 3000+, since they have the highest clock speed for Socket A Semprons (except the Athlon XP\'s, of course)... At first I thought of upgrading with an AthlonXP 3200+, but then I saw that I\'ll get a far more advanced cpu and mobo for almost the same price (XP 3200+ are sold at about 140Euros!).

And you\'re a 100% correct, every little bit helps (not only in theory, as BOINC shows!)


My NEW BOINC-Site

Why people joined BOINC Synergy...
ID: 27847 · Report as offensive     Reply Quote
Profile astroWX
Volunteer moderator

Send message
Joined: 5 Aug 04
Posts: 1496
Credit: 95,522,203
RAC: 0
Message 27848 - Posted: 12 Apr 2007, 18:22:58 UTC
Last modified: 12 Apr 2007, 18:25:13 UTC

C/L is Command Line. If it can\'t use X, then you can\'t see the \"Show Graphics\" option. I presume you can see the boinc log; there\'s a list of one-line entries (in full screen) with RunID, time step, and run time information. These are logged at each Checkpoint and Checkpoints occur every 432 Time Steps. You\'d lose the least amount of work by exiting immediately after an entry is made.

If you can\'t see the log, you\'re flying blind and each shut-down will cost, on average, 216 Time Steps to rerun.
"We have met the enemy and he is us." -- Pogo
Greetings from coastal Washington state, the scenic US Pacific Northwest.
ID: 27848 · Report as offensive     Reply Quote
old_user29078

Send message
Joined: 9 Nov 04
Posts: 11
Credit: 93,753
RAC: 0
Message 27850 - Posted: 12 Apr 2007, 19:15:00 UTC - in response to Message 27848.  

C/L is Command Line. If it can\'t use X, then you can\'t see the \"Show Graphics\" option. I presume you can see the boinc log; there\'s a list of one-line entries (in full screen) with RunID, time step, and run time information. These are logged at each Checkpoint and Checkpoints occur every 432 Time Steps. You\'d lose the least amount of work by exiting immediately after an entry is made.

If you can\'t see the log, you\'re flying blind and each shut-down will cost, on average, 216 Time Steps to rerun.


Thanks. Do you know where I\'d find the log?

I\'ve had a poke around in my boinc folder, but the only log I can find is simply a copy of the \"Messages\" tab of BOINC Manager and doesn\'t seem to include any specific logging information for CPDN.

ID: 27850 · Report as offensive     Reply Quote
Profile old_user206071
Avatar

Send message
Joined: 29 Oct 06
Posts: 14
Credit: 99,628
RAC: 0
Message 27851 - Posted: 12 Apr 2007, 19:46:26 UTC - in response to Message 27850.  

Thanks. Do you know where I\'d find the log?

I\'ve had a poke around in my boinc folder, but the only log I can find is simply a copy of the \"Messages\" tab of BOINC Manager and doesn\'t seem to include any specific logging information for CPDN.


I dont know where to find the log, but you can edit the cc_config.xml to get more informations from BOINC (not sure if it works under Linux, though). I did that to see when SAP/CPDN last checkpointed...

cc_config.xml:

<cc_config>

<log_flags>

<task_debug>1</task_debug>

</log_flags>

</cc_config>

Then in stdoutdae.txt it saves additional info like this:

2007-04-12 21:42:33 [climateprediction.net] [task_debug] result hadcm3inct_cm3j_1920_160_05863680_0 checkpointed


My NEW BOINC-Site

Why people joined BOINC Synergy...
ID: 27851 · Report as offensive     Reply Quote
Profile MikeMarsUK
Volunteer moderator
Avatar

Send message
Joined: 13 Jan 06
Posts: 1498
Credit: 15,613,038
RAC: 0
Message 27852 - Posted: 12 Apr 2007, 19:48:58 UTC


I do this also, very useful. Note that this works with 5.8.15, but for 5.8.16 I think you need <checkpoint_debug>1</checkpoint_debug> instead of task_debug.

I'm a volunteer and my views are my own.
News and Announcements and FAQ
ID: 27852 · Report as offensive     Reply Quote
Profile old_user206071
Avatar

Send message
Joined: 29 Oct 06
Posts: 14
Credit: 99,628
RAC: 0
Message 27853 - Posted: 12 Apr 2007, 19:55:27 UTC - in response to Message 27852.  


I do this also, very useful. Note that this works with 5.8.15, but for 5.8.16 I think you need <checkpoint_debug>1</checkpoint_debug> instead of task_debug.


I\'m using 5.8.16 under WinDoze, it still works! ;)


My NEW BOINC-Site

Why people joined BOINC Synergy...
ID: 27853 · Report as offensive     Reply Quote
old_user29078

Send message
Joined: 9 Nov 04
Posts: 11
Credit: 93,753
RAC: 0
Message 27854 - Posted: 12 Apr 2007, 20:49:20 UTC - in response to Message 27851.  

I can\'t find either cc_config.xml or stdoutdae.txt anywhere on my system, so I guess I\'ll just have to continue as I have been for the time being.

Mind you, I see these posts refer to a newer version of BOINC than I\'m running. However, a new OS release is due in a few days\' time, so hopefully I will be more up to date and may have access to extra features then.

ID: 27854 · Report as offensive     Reply Quote
Profile old_user206071
Avatar

Send message
Joined: 29 Oct 06
Posts: 14
Credit: 99,628
RAC: 0
Message 27855 - Posted: 12 Apr 2007, 21:07:00 UTC - in response to Message 27854.  

I can\'t find either cc_config.xml or stdoutdae.txt anywhere on my system, so I guess I\'ll just have to continue as I have been for the time being.

Mind you, I see these posts refer to a newer version of BOINC than I\'m running. However, a new OS release is due in a few days\' time, so hopefully I will be more up to date and may have access to extra features then.


I asked at our team site if there is a linux user that could help you... Cant promise, though...


My NEW BOINC-Site

Why people joined BOINC Synergy...
ID: 27855 · Report as offensive     Reply Quote
Profile MikeMarsUK
Volunteer moderator
Avatar

Send message
Joined: 13 Jan 06
Posts: 1498
Credit: 15,613,038
RAC: 0
Message 27856 - Posted: 12 Apr 2007, 21:33:56 UTC


You\'d need to create cc_config.xml yourself - Ageless\'s FAQs describe how to do it (PS it\'s worth keeping a shortcut to them).

But cc_config only works with 5.8.* of the manager, not earlier versions.

http://boincfaq.mundayweb.com/index.php

If your system can display the project graphics, there\'s another way to find out when the next checkpoint is : \'Show Graphics\', \'Z\' (to hide BBC sidebar if it\'s on the left), \'8\' (to display countdown timer to the next checkpoint).
I'm a volunteer and my views are my own.
News and Announcements and FAQ
ID: 27856 · Report as offensive     Reply Quote
Kenneth Larsen

Send message
Joined: 26 Aug 04
Posts: 59
Credit: 438,133
RAC: 0
Message 27864 - Posted: 13 Apr 2007, 8:03:08 UTC - in response to Message 27854.  

I can\'t find either cc_config.xml or stdoutdae.txt anywhere on my system, so I guess I\'ll just have to continue as I have been for the time being.


Hi, I saw Shauhulud´s request for help at our team site.
To search for the mentioned files, the following command should help you:
find / -iname \'cc_config.xml\' -or -iname \'stdoutdae.txt\'

This will search for the file from the root drive and up, meaning all libraries.

If you already have performed a full search, I probably can\'t help you :-(
Proud owner of the CPDN Wow-Mug!
ID: 27864 · Report as offensive     Reply Quote
Jord
Avatar

Send message
Joined: 5 Aug 04
Posts: 250
Credit: 93,274
RAC: 0
Message 27877 - Posted: 13 Apr 2007, 18:06:20 UTC - in response to Message 27856.  
Last modified: 13 Apr 2007, 18:11:36 UTC

cc_config.xml must be made by the person operating the computer. It isn\'t made by BOINC. So you don\'t need to go search for it if you never made it in the first place. :-)

stdoutdae.txt is made by BOINC though. It\'s the text file that holds the log of messages you see in Boinc Manager\'s Messages tab. Other files used to store these messages are stderrdae.txt (for specific errors), stderrgui.txt (for GUI specific errors) and stdoutgui.txt (for trace errors).

But cc_config only works with 5.8.* of the manager, not earlier versions.

Rectification: It does work with earlier versions of BOINC. It\'s just that not all commands available now work on older versions. I still want to find which commands were available on older versions, but can\'t find the time to do so. ;)
Jord.
ID: 27877 · Report as offensive     Reply Quote

Message boards : Number crunching : Unable to complete WU by deadline: should I continue to participate?

©2024 climateprediction.net