| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Support
Forum: Website Support
Thread: Run time on stats page is much lower than actual run time |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 5
|
|
| Author |
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hi,
I'm not sure if this is the right forum, but I think this fits here at best. according to my stats page i have run Smash Childhood Cancer for 0:003:18:12:36 or roughly 90hrs. as it happens I had running WUProp all the time. When you look at my profile@WUProp then you can see that I have a run time of 164 hrs. thats a huge difference. so, at first I thought I had a lots of open WUs. But I only have 5 pending validation. that less then 10hrs. thats strange... maybe it was my Laptop? My android device is running OpenZika for 0:009:19:37:59 or roughly 235hrs. Again I looked at WUProp and there I see 485 hrs (20+ days!). 2 WUs are pending, thats 16hrs... That would be the 14day badge. :( so, I did some testing with primegrid, einstein and universe. It looks like WUProp is really accurate in terms of run time. so I looked closer. it could be run time vs cpu time. BUT, that works only on my android. according to boinc boinc is allowed to compute 92.07% of the time my laptop is running boinc. math: 164 *,92 = 150 --> 150 - 10 = 140 --> 140 > 90 according to WCG and some math my SCC WUs have an 8% longer run time then CPU time...so that looks okay. but there is still some time missing... I did the same for my android. boinc says: 79.41% of the time boinc is runnig my andorid is allow to compute. I did some math with my WUs. cpu time vs run time is 52%...that won't work...not even close. but WUProp also takes WUs into account, that fail or are not verified. Again, I checked my results, but I only found 1 WUs which status is pending verfification. But I have no failed, error, aborted, too late or similar WUs. At least not in the last 5 days. (thats more or less the time scince I run SCC) I run OpenZika much longer...so no Idea if there where a lot of WUs which did not validate.... so, the only thing that could be, is that WCG stats are to much behind. today I did 35hrs in scc (until now) but not since 1200 UTC (maybe 18hrs since 1200 UTC)... and with OpenZika it would be...2 core in my andorid...ahm...not sure, but much less^^ so, the questions is: where is my run time? please enlighten me :( |
||
|
|
mmonnin
Advanced Cruncher Joined: Jul 20, 2016 Post Count: 148 Status: Offline Project Badges: 
|
Isn't WCG CPU time? I think I tested running 2 clients on one PC. I didn't get twice the time since there were 2 clients each taking ~50% CPU time. WUProp is run time.
---------------------------------------- |
||
|
|
PMH_UK
Veteran Cruncher UK Joined: Apr 26, 2007 Post Count: 786 Status: Offline Project Badges: 
|
Isn't WCG CPU time? I think I tested running 2 clients on one PC. I didn't get twice the time since there were 2 clients each taking ~50% CPU time. WUProp is run time. I believe this is correct, WUprop measures elapsed time but WCG measures CPU time. One of my PCs thermal limits CPU with TThrottle and it revealed this. Paul.
Paul.
|
||
|
|
Dark Angel
Veteran Cruncher Australia Joined: Nov 11, 2005 Post Count: 728 Status: Offline Project Badges: 
|
The actual amount of CPU time is affected by various things. Memory bandwidth, which I mean to include CPU cache at various levels as well as system is the first bottleneck where a CPU can be made to wait fractions of a second every time it wants data. The only thing you can change here is to replace the RAM with faster modules IF your motherboard supports them. Next is the main storage system ie the hard drive. Some projects have seen significant improvements in efficiency (CPU time vs run time) when switched to a SSD over a spinning platter HDD. This is probably the biggest improvement in system efficiency you can do.
----------------------------------------Both areas are affected by other applications running on the system. The more you use you PC the more system resources are diverted to user applications rather than to BOINC. BOINC runs at a lower system priority so it gives up resources, both CPU and RAM, when they are required by user programs. This will have a big impact on efficiency if you are doing something that causes a lot of HDD activity or uses a lot of CPU cycles and system RAM. To improve the efficiency of the system without investing in better hardware you need to disable competing processes, especially those that use the HDD. Desktop visual effects, file indexing, virus scanning the BOINC folder, these use up system resources and reduce the available CPU time for your BOINC tasks.  Currently being moderated under false pretences |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Isn't WCG CPU time? I think I tested running 2 clients on one PC. I didn't get twice the time since there were 2 clients each taking ~50% CPU time. WUProp is run time. I ruled that out above...run time multiplied with the computing time factor of boinc should be cpu time. that works fine when I look at my WUs in WCG, but I'm missing still 20+ hrs. btw, run time multiplied with the computing time factor of boinc is almost exactly the cpu time for other projects, (tested at einstein, primegrid and universe) but I will run a test batch of 6 hours tomorrow. maybe there is something else...I only measured single WUs till now. @Dark Angel: I'm runnig Bonic on my laptop on a Samsung 960 (M2/M-Key, NVMe) Pro. Not sure what could be really faster. ^^ Also I have 16 GB RAM, I can't really switch that...but at peak there are 4GB used by boinc and only ~7GB are used in total...so I see no reason for a memory limit... tbh at first I thougt running primgrid (PG is really consuming ram bandwith) could limit something. but I stopped it a few days ago and nothing changed... something completly different regarding SCC stats: I have 38 results returned in 90 hours. thats ~2,37hrs per result. (avarage) my longst running result is 2,37hrs (cpu time) / 2,46 (running time). second is 2,27 / 2,40, third is 2,18 / 2,29. the shortests was 1,08 / 1,12. not sure, but my knowledge of statistics says thats not possible.  looking at my results I would have assumed avargae is 1,5 or 1,6hrs...maybe 1,7 hrs[Edit 2 times, last edit by Former Member at Dec 17, 2018 11:31:07 PM] |
||
|
|
|