| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Completed Research
Forum: Computing for Clean Water Forum
Thread: Long recalculated times to completion |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 9
|
|
| Author |
|
|
Baywhale
Advanced Cruncher Joined: Apr 18, 2006 Post Count: 88 Status: Offline Project Badges: 
|
I run CCW on my desktop and laptop. On the (less powerful) laptop BOINC often recalculates the time to completion to such a high number of hours it makes more sense to abort the task.
----------------------------------------On my laptop, typically the initial time to completion is around 20 hours but when the task is almost complete it can leap to 60 hours with only a few days left to run the task — even when there is no possibility of completion I leave the task running as it sometimes reverts to the original 20-hour figure. Aborting days worth of work is obviously wasteful — is there a simple fix to the problem or would I be better to run a different project altogether on the laptop? |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Baywhale,
----------------------------------------Is your BOINC client set to only use a % of the CPU time? My laptop always takes about 6.6 CPU hours for a CFCW job, but depending how busy it is, the Elapsed time can be 7 to 8.5 hours. That is the projected time which sometimes can show very inflated values. You don't say what client u have installed, but assuming 6.10.58/59 click on the running task in BOINC Manager and then the properties button at left which gives the CPU time also. Long at that keeps increasing and the last checkpoint info updates once in a while things are fine and save to leave the task running. Clean Water normally has a deadline of 10 days. Do you keep allot of tasks cached, because your "with only a few days left to run the task" suggests that, unless those are repair jobs that have a 4 day deadline. That means your computer is rated by WCG as reliable, returning valid jobs on a continuous basis. --//-- [Edit 1 times, last edit by Former Member at Jul 16, 2011 11:49:45 AM] |
||
|
|
Baywhale
Advanced Cruncher Joined: Apr 18, 2006 Post Count: 88 Status: Offline Project Badges:
|
I run BOINC on my laptop at 65% of CPU to avoid over-heating problems, and clearly as a result this extends the time that jobs take to compute.
----------------------------------------I tend to keep my additional work buffer to a minimum (it's set at 0.10 at the moment). The laptop isn't switched on continuously so it does need the best part of the ten day deadline to complete a 20-30 hour task — the sudden leap to 60 hours makes it impossible to reach within the 10 days. The client is 6.10.58 and the task properties do show that the task continues proceeding. The problem is these leaps in the length of time to completion. I haven't come across this problem with other projects. Is there an aspect of CCW that revises the projected time to completion so radically when the job appears close to finishing? |
||
|
|
sk..
Master Cruncher http://s17.rimg.info/ccb5d62bd3e856cc0d1df9b0ee2f7f6a.gif Joined: Mar 22, 2007 Post Count: 2324 Status: Offline Project Badges: 
|
65% will use one core from a dual core...
Can't answer your other questions, but on other projects the ET of completion varies due to several projects with different run times, and Boinc seeing each project as one Project (from the same grid). Any system crashes, restarts? If not this might be of some interest to a tech. IMO the best WCG project for lesser systems is HCMD2. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
I run BOINC on my laptop at 65% of CPU to avoid over-heating problems, and clearly as a result this extends the time that jobs take to compute. I tend to keep my additional work buffer to a minimum (it's set at 0.10 at the moment). The laptop isn't switched on continuously so it does need the best part of the ten day deadline to complete a 20-30 hour task — the sudden leap to 60 hours makes it impossible to reach within the 10 days. The client is 6.10.58 and the task properties do show that the task continues proceeding. The problem is these leaps in the length of time to completion. I haven't come across this problem with other projects. Is there an aspect of CCW that revises the projected time to completion so radically when the job appears close to finishing? Propose to look at the regular [quite constant] reported CPU time on the Result Status page for a C4CW task and compare that to the CPU time ** of a task in progress. When that is say 6 hours of the regular 7, you got a good indication of the real progress. C4CW does regularly save progress to disk, so if you interrupt the calculations, there should be little loss on power-up, but it is important to power down properly. Best use the laptop hibernation function, then on resume there's normally not a second of progress loss. If the laptop has been off for longer then yes the client recomputes the projected completion time based on recent progress/computing history, but it is wall-clock time, not true computing [CPU] time! Generally, I'd not worry about it too much. Maybe sometimes let the laptop run over-night to catch up and all will be fine. --//-- edit: ** In the Task properties screen. [Edit 1 times, last edit by Former Member at Jul 17, 2011 7:55:53 AM] |
||
|
|
Baywhale
Advanced Cruncher Joined: Apr 18, 2006 Post Count: 88 Status: Offline Project Badges:
|
Thanks for the replies. I'm going to pay more attention to when this problem occurs — I think powerdowns might be where the problem lies. If I can't work round it then I'll try HCMD2.
---------------------------------------- |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
When you get the inflated times it means that the check point file
----------------------------------------has been corrupted. As a result the work unit will not be valid when it completes. Any time you kick it out of memory this is a possibility. [Edit 1 times, last edit by Former Member at Jul 18, 2011 2:12:22 AM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Anyone else seeing such results going ´error´ or ´invalid´?
--//-- |
||
|
|
joeperry39@gmail.com
Advanced Cruncher USA Joined: Nov 22, 2006 Post Count: 140 Status: Offline Project Badges: 
|
Anyone else seeing such results going ´error´ or ´invalid´? --//-- I'm not getting "error" or "invalid," however I do seem to recently have a much larger number of "inconclusive" results. They seem to run in bunches: go several days with none, then several at a time. edit: Also several "awaiting validation" that take a few days to run thru the validation process.  "Everything in moderation, including moderation" -- Mark Twain [Edit 1 times, last edit by osugrad at Jul 18, 2011 11:15:06 PM] |
||
|
|
|