| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
| World Community Grid Forums
|
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 10
|
|
| Author |
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
After upgrading my Boinc device (Vista 32b, 2 cores) a few days ago, i've been experiencing all 3 of my uploaded work units validated with less CPU time as they were showing before uploading.
----------------------------------------faah13700_ ZINC11628124_ xEyeSiteXtl5NI_ 00_ 1 -- Pending Validation -- 05.10.10 14:41:11 06.10.10 14:25:29 12.59 115.9 / 0.0 --> actually ~13.5h CPU time faah13680_ ZINC05933874_ xEyeSiteXtl5NI_ 00_ 0-- In Progress 04.10.10 19:58:19 14.10.10 19:58:19 0.00 0.0 / 0.0 faah13680_ ZINC05751933_ xEyeSiteXtl5NI_ 00_ 0-- Valid 04.10.10 19:58:01 06.10.10 08:28:01 13.98 128.3 / 105.4 --> actually ~15h CPU time faah13653_ ZINC00129835_ xEyeSiteXtl5NI_ 01_ 0-- Valid 03.10.10 21:47:07 05.10.10 22:34:20 16.63 155.0 / 113.1 --> actually ~24.5h CPU time As the upgrade did't work out smoothly, i had to manually delete my old WCG-Boinc. After losing 8h of CPU time in the third example shown above, i tried the standard (non-WCG) Berkeley Boinc, but with the same results. After that i reistalled the new WCG-Boinc again. I suspected, that overriding the Boinc-Data directory with my old data dir (with unfinished wu's) might have caused the problem, however, the latest result shown above had been downloaded after overriding the Boinc data directory. I'm now letting the last result finish and then i make a completely fresh install without porting any old data. I had only FAAH units running since the upgrade, and all 3 of them had been validated with less CPU time. Any ideas what might be the reason? [Edit 1 times, last edit by Former Member at Oct 7, 2010 4:23:16 PM] |
||
|
|
GB033533
Senior Cruncher UK Joined: Dec 8, 2004 Post Count: 206 Status: Offline Project Badges:
|
Yeah, I've noticed that too! It seems to be something to do with the BOINC manager showing 'Elapsed' time, but you only get credited for CPU time. You can see the difference between the two when you look at the properties while it's running, but not once it's completed.
----------------------------------------Maybe one of the community advisors can elaborate.... ![]() |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
so here's the last one:
faah13680_ ZINC05933874_ xEyeSiteXtl5NI_ 00_ 0: 14.80 hrs boinc manager showed: elapsed: 15:37:51 i don't know if the new version shows elapsed real time (wall time) all of a sudden, but prior versions showed the CPU time, which was exactly the same time you get granted then. so i've completely re-installed boinc with a new data dir by now and i'll try some non-faah as well, as soon as the project servers get back on-line ;) |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
Yes, there's been an FAQ for longer on this change to showing "Elapsed" (Wallclock time when BOINC was allowed to run) instead of CPU time in anticipation of what was coming with 6.10. Essentially this shows that your system was doing other things for about 50 minutes out of the 15:37 hours. It's always done that except now volunteers notice. Depending on how 'heavy' a science is, this differential can be larger or smaller. Generally my systems hit 99.5-99.9% efficiency during the night and about 90-95% during use.
----------------------------------------
WCG
Please help to make the Forums an enjoyable experience for All! |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
thanks Sekerob, so the mystery is solved.
well, so the good ol' cpu clock is gone... i liked it better that way :( for a standard user, however, how runs boinc at 75% cpu usage and suspended at 25% system usage, the wallclock should differ way more. personally, i would have liked an clock showing "last restore point... mins", but that's another story. thanks for helping :) |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
If on windows, get BOINCTasks BOINC Manager replacement. It shows both Elapsed and CPU time AND time since last checkpoint. It's all I use these days, plus it allows to watch all systems in a lan, Linux and Macs inclusive from 1 Windows computer or virtual machine.
----------------------------------------
WCG
Please help to make the Forums an enjoyable experience for All! |
||
|
|
ski1939
Senior Cruncher Freeport, Illinois, USA Joined: Nov 17, 2004 Post Count: 209 Status: Offline Project Badges:
|
Sekerob - do you have a link for the "BOINCTasks BOINC Manager replacement"?
---------------------------------------- |
||
|
|
Dataman
Ace Cruncher Joined: Nov 16, 2004 Post Count: 4865 Status: Offline Project Badges:
|
Sekerob - do you have a link for the "BOINCTasks BOINC Manager replacement"? I'm not Sekerob but the latest versions I know of are here: http://www.efmer.eu/boinc/boinc_tasks/download.html ![]() |
||
|
|
kateiacy
Veteran Cruncher USA Joined: Jan 23, 2010 Post Count: 1027 Status: Offline Project Badges:
|
thanks Sekerob, so the mystery is solved. for a standard user, however, how runs boinc at 75% cpu usage and suspended at 25% system usage, the wallclock should differ way more. personally, i would have liked an clock showing "last restore point... mins", but that's another story. For your standard user, it's even more confusing than that, because the "elapsed" time isn't truly wallclock time. "Elapsed" time seems to be the amount of time that BOINC has been able to request CPU time for the WU. So for a person running at 75% CPU usage, the elapsed time shown is actually only 3/4 as long as the WU has already been running, and the "time to completion" is only 3/4 as long as the real estimated wallclock time. The CPU time still will be shorter than the shown "elapsed" time, because if the computer is doing other things, the WU won't actually get and use all the CPU time it's permitted to ask for. I think it would be nice if "elapsed" really meant "elapsed" -- i.e., wallclock -- even when a machine wasn't set for 100% cpu usage. I run my laptop at 65% CPU usage for heat reasons. I don't want to reboot it while a CEP2 task is running because of the infrequent benchmarks. It would be convenient not to have to do mental math to figure out the actual estimated time to completion. Obviously not a big deal, however :) ![]() |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Thanks for all your advices, BoincTasks sounds like a useful tool, i'll check it out.
[ot] kateiacy, if you have problems with overheating, try tthrottle (same link as BoincTasks above). i've been using this one for several months now. |
||
|
|
|