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: 8
|
![]() |
Author |
|
retsof
Former Community Advisor USA Joined: Jul 31, 2005 Post Count: 6824 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Most of the time this computer runs 100% WCG and nothing else.
----------------------------------------I usually run other BOINC projects here and there for a bit of testing, just to see what is going on out there. DIMES and grid.org and the UD version of WCG are available, but exited at the moment. Several BOINC projects have nice BOINC graphics, and Rosetta@home has especially useful stick figure folding and energy line real-time graphics that could be applied here if Phase II of our HPF project ever resumes. It's real folding, not just a molecule animation. I enabled several other projects on one computer to see how the current BOINC scheduler is working. It usually runs one project until it panics and sees something that needs work to complete before the completion date. That's regardless of the requested resource share. I have opened everything up, with the exception of three climate simulation projects that take months to complete a workunit. Those are beautiful works of dynamic weather, temperature, cloud simulation art, etc., but should be run in standalone mode with no other projects. Checkpoints are far apart, and project splitting didn't work well when those were active. Even giving WCG the major part of the allocation does not seem to help so far. It has 76.92% of this computer, but 10 workunits ran to completion and have so far not been replaced. (why not, given the major allocation? It may have something to do with the mix of reporting dates and estimated execution times.) Here's what my schedule competion computer looks like at the moment, with QMC@home currently running. It's 3 hours into a 12 hour estimated workunit. The BOINC graphic shows a couple of stick-and-ball molecules with generated points for electron position. Some other project workunits complete in as little as 2 seconds. Each one of these also contributes to a different Vulture Central team. (listed at the end of the Vulture Central team thread) Project - Resource Share CPDN Seasonal Attribution - 5 (0.77%) empty, won't get new tasks BBC Climate Change Experiment - 5 (0.77%) empty, won't get new tasks Rosetta@home - 5 (0.77%) HashClash - 5 (0.77%) BoincSimap - 5 (0.77%) Leiden Classical - 5 (0.77%) Riesel Sieve - 5 (0.77%) vtu@home - 5 (0.77%) BURP - 5 (0.77%) Rectilinear Crossing Numbers - 5 (0.77%) 1 preempted after 5 seconds, 11 workunits waiting Einstein@home - 5 (0.77%) 1 preempted after 4 hours, 47 minutes Project TANPAKU - 5 (0.77%) lhcathome - 5 (0.77%) Project Neuron - 5 (0.77%) Pirates@home - 5 (0.77%) Predictor@home - 5 (0.77%) QMC@home - 5 (0.77%) running, 3 hours 45 minutes @ 28% ralph@home - 5 (0.77%) Spinhenge@home - 5 (0.77%) 3 workunits waiting SZTAKI Desktop Grid - 5 (0.77%) 1 workunit waiting chess960@home alpha - 5 (0.77%) NanoHive@home - 5 (0.77%) PrimeGrid - 5 (0.77%) uFluids - 5 (0.77%) 1 workunit waiting Worldcommunitygrid - 500 (76.92%) XtremLab - 5 (0.77%) Climateprediction.net - 5 (0.77%) empty, won't get new tasks SETI@home beta test - 5 (0.77%) 1 workunit waiting SETI@home - 5 (0.77%) RenderFarm@home - 5 (0.77%) crashcollection - 5 (0.77%) no tasks, no points, no execution, no teams
SUPPORT ADVISOR
----------------------------------------Work+GPU i7 8700 12threads School i7 4770 8threads Default+GPU Ryzen 7 3700X 16threads Ryzen 7 3800X 16 threads Ryzen 9 3900X 24threads Home i7 3540M 4threads50% [Edit 4 times, last edit by retsof at Sep 28, 2006 6:38:58 PM] |
||
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
Hi Retsof,
----------------------------------------u might want to visit the relevant work scheduler and buffering pages at the unofficial BOINC wiki. They provide samples of how to determine the switch frequency, contact frequency and explanation for the scheming of Long Term Debt, Short Term Debt, Earliest Time Due, Panic etc. Got 4 projects running and it works 'almost' flawless and pulls across work on a trickle basis just enough to keep the buffer full. It takes quite a while for it to balance all and becomes extra problematic because of the highly variable lengths of the WCG WU's....HDC's from 0.75 to 2 hours, FAAH from 7 to 8.5 hours and the ever recomputed Time To Complete projection of the WU's in queue for ech project separately. In BOINCview u can actually see the LTD/STD times for all the projects which gives u the lead how each overall is doing. Mixing WU's of seconds with those of 8 hours is a serious curve ball, but in 5.6.3 it's doing a pretty good job. ciao
WCG
----------------------------------------Please help to make the Forums an enjoyable experience for All! [Edit 1 times, last edit by Sekerob at Sep 28, 2006 6:39:28 PM] |
||
|
retsof
Former Community Advisor USA Joined: Jul 31, 2005 Post Count: 6824 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Mixing WU's of seconds with those of 8 hours is a serious curve ball, but in 5.6.3 it's doing a pretty good job. curve ball .... fast ball .... fast ball .... fork ball .... curve ball .... spit ball .... pitcher ejected from gameciao
SUPPORT ADVISOR
Work+GPU i7 8700 12threads School i7 4770 8threads Default+GPU Ryzen 7 3700X 16threads Ryzen 7 3800X 16 threads Ryzen 9 3900X 24threads Home i7 3540M 4threads50% |
||
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
must be a merkenism like the bottom eight.....woosh
----------------------------------------BTW noticed on the last series of HDC and FAAH thru BOINC 5.6.3, that there is no x-type TTC adjusting i.e. when HDC's complete it's not adjusting the FAAH's which should improve scheduling considerably.
WCG
----------------------------------------Please help to make the Forums an enjoyable experience for All! [Edit 1 times, last edit by Sekerob at Sep 29, 2006 6:16:07 AM] |
||
|
retsof
Former Community Advisor USA Joined: Jul 31, 2005 Post Count: 6824 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
The multi-project computer is still running a mixture of 5 minute through 8 hour projects. NO WORLDCOMMUNITYGRID projects have been downloaded in several days even though it is set to 76% of the available resources. WCG must not work and play well with others. Is something wrong with WCG scheduling in combination with other BOINC projects? I know it works if ONLY WCG is set or if other projects are suspended to force a WCG download.
----------------------------------------The project rotation has gone around a few times with due dates ranging from 12 hours after download through October 21. Rectilinear Crossing - 1 preempted, 11 waiting SZTAKI - 1 waiting uFluids - 1 waiting Spinhenge - 1 preempted, 2 waiting Einstein@home - 1 preempted Seti@home beta - 1 waiting QMC@home - 1 preempted Riesel Sieve - 1 preempted vtu@home - 1 waiting (2 had computation errors, computer is not overclocked, other project workunits running correctly) HashClash - several downloaded/uploaded Leiden Classical - running, 5 waiting Several chess960@home and xtremlab workunits have run in addition to others from the above projects. I am not attempting to count every workunit. I am watching the current mixture. In the meantime, Leiden Classical finished and some NEW chess960@home and extremlab workunits downloaded and chess960@home is running. Some other projects are down or have no work available. If I get no WCG workunits after a week or two I will set the other projects to No New Tasks so that they will finish and WCG will resume. There are not very many projects in the queue, perhaps due to a screwy estimate from QMC@home. It has an estimate of 225 hours, but it is now at 8 hours out of an estimated 12-14 hours. So far, this may be an indication that WCG is underutilized if multiple projects are running.
SUPPORT ADVISOR
----------------------------------------Work+GPU i7 8700 12threads School i7 4770 8threads Default+GPU Ryzen 7 3700X 16threads Ryzen 7 3800X 16 threads Ryzen 9 3900X 24threads Home i7 3540M 4threads50% [Edit 5 times, last edit by retsof at Sep 30, 2006 1:44:07 AM] |
||
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
Already mentioned BOINCview that can show all the Long/Short terms (Time) debts for all to show in the Projects Tab (add hidden columns). If u think it's out of whack, look at this utility for Windows XP
----------------------------------------BOINC Debt Viewer It allows to reset the values so it starts the balancing act from scratch. Also follow this discussion, how to reset the values manually. http://boincview.amanheis.de/forum/viewtopic.php?t=461 bona fortuna PS 5.6.4 has even more refinement in the scheduling.....Last FAAH processed slaps it's CPU completion time onto any in work buffer, not impacting the HDC's.
WCG
Please help to make the Forums an enjoyable experience for All! |
||
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
With all the tasks ready to run for the low priority projects the client is most likely unable to download anything else without getting itself (farther) into deadline trouble. If you leave it be once it clears out all the work from the low priority projects you will not hear from them again for a long time and will only have WCG work most of the time.
http://boinc-wiki.ath.cx/index.php?title=BOINC_Powered_Backup_Project See this wiki page for more info on the expected behavior when running projects with significant differences in resource share. |
||
|
retsof
Former Community Advisor USA Joined: Jul 31, 2005 Post Count: 6824 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
I think the problem resolved itself. The excessively overestimated QMC@home workunit ran one more hour overnight, from 8 hours to 9 hours. The time to completion estimate dropped from 255 hours to 191 hours. That freed up some time. There were 10 WCG FAAH workunits waiting this morning, which will take about 50 hours to run. (I am only running FAAH)
----------------------------------------BAAAAAD QMC. No more tasks for you.
SUPPORT ADVISOR
----------------------------------------Work+GPU i7 8700 12threads School i7 4770 8threads Default+GPU Ryzen 7 3700X 16threads Ryzen 7 3800X 16 threads Ryzen 9 3900X 24threads Home i7 3540M 4threads50% [Edit 2 times, last edit by retsof at Sep 30, 2006 1:22:09 PM] |
||
|
|
![]() |