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 |
|
Datanalle
Cruncher Joined: Oct 9, 2015 Post Count: 3 Status: Offline |
I've been using the worldcommunitygrid.org BOINC client at 50-90% CPU utilization (usually using half of the cores) since 2015 without ever having any performance issues. In fact around January (after upgrading my hardware to this decade) to May 2017 by some black magic and high-tech witchcraft I was able to play AAA games while simultaneously running BOINC at ~80% CPU utilization with 3/4 cores and GPU computing enabled. I've usually disabled BOINC while gaming but after doing it once by accident and not noticing any performance drops, I decided to leave BOINC on regardless of what I was doing. I was churning out a solid 10 000 points a day on average with one or two dozen returned results and all was well.
Sometime around late June or early July 2017 something happened that ruined my ability to run this program at any meaningful level. Running BOINC at my previous CPU utilization levels made the computer a massive pain to use as the OS would frequently freeze for a fraction of a second multiple times a minute. These freezes stop video and audio playback as well as make the system completely unresponsive. Although some days the system ran just fine for a whole day. Nowadays I've settled for running at 20% CPU utilization with 2/4 cores and I still get at least a couple of 0.1-3.0 second long freezes a day. I've had to turn off BOINC at times when watching YouTube videos has gotten too frustrating. These days I generate around 1 000 points a day with ~3 returned results. My hardware has not changed since January 2017 and there have been no meaningful changes to my software environment around the time these freezes started. As far as I'm concerned they started completely randomly. I have recently tried using the client from worldcommunitygrid.org as well as from boinc.berkeley.edu but both have the same issue. Back in 2016 when I tried the Berkeley client it worked fine. This problem, however, is intermittent. Right now I am running with 99% utilization with 4/4 cores, I've manually changed the priority of the 4 running processes (tasks currently computing) to "high" with the Task Manager (they are "low" by default), my CPU usage is pegged at 100% and the system is working fine. Temperatures are not an issue with my setup and the freezes are not related to how long I've had the system on for. I am certain the issue is with this program because the freezes stop within a second or two of closing the client. I am running Windows 7 64-bit and I have primarily only contributed to the Mapping Cancer Markers project so the issue has nothing to do with a specific project. |
||
|
mmonnin
Advanced Cruncher Joined: Jul 20, 2016 Post Count: 148 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Do you always run with a High priority? They should default to the lowest priority. You mentioned GPU crunching. For which project? Some will cause video/input lag at default settings, others do not.
----------------------------------------![]() |
||
|
Datanalle
Cruncher Joined: Oct 9, 2015 Post Count: 3 Status: Offline |
Do you always run with a High priority? They should default to the lowest priority. You mentioned GPU crunching. For which project? Some will cause video/input lag at default settings, others do not. No, by default the processes run at the lowest priority. I was simply running them at high for a day to test if that would immediately cause freezing, but it didn't. I have "Use GPU always" enabled in the activity menu and I'm currently only working on the Mapping Cancer Markers project. I don't know if there's a way to select which projects use the GPU. I've suspended GPU usage from the activity menu and web settings for now to see if that changes anything. |
||
|
KLiK
Master Cruncher Croatia Joined: Nov 13, 2006 Post Count: 3108 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
1st of all, tell us more about your equipment. Is it a laptop or a PC? What configuration do you run?
----------------------------------------2nd, you should also run WCG temperature regulated. For example, I useTthrottle , but some others might have alternative options. ![]() |
||
|
Sgt.Joe
Ace Cruncher USA Joined: Jul 4, 2006 Post Count: 7697 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
I would check for a rogue process using task manager or process lasso. Check to see if some process is accruing time when it should not be. A long time ago I had a svchost.exe which was accruing time which I think was caused by a botched update, but it was so long ago I can't recall the details. Another possibility is your antivirus maybe checking too often or too aggressively.
----------------------------------------Cheers
Sgt. Joe
*Minnesota Crunchers* |
||
|
mmonnin
Advanced Cruncher Joined: Jul 20, 2016 Post Count: 148 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Do you always run with a High priority? They should default to the lowest priority. You mentioned GPU crunching. For which project? Some will cause video/input lag at default settings, others do not. No, by default the processes run at the lowest priority. I was simply running them at high for a day to test if that would immediately cause freezing, but it didn't. I have "Use GPU always" enabled in the activity menu and I'm currently only working on the Mapping Cancer Markers project. I don't know if there's a way to select which projects use the GPU. I've suspended GPU usage from the activity menu and web settings for now to see if that changes anything. Well WCG is CPU only so the GPU settings make no difference. ![]() |
||
|
SekeRob
Master Cruncher Joined: Jan 7, 2013 Post Count: 2741 Status: Offline |
Do you always run with a High priority? They should default to the lowest priority. You mentioned GPU crunching. For which project? Some will cause video/input lag at default settings, others do not. No, by default the processes run at the lowest priority. I was simply running them at high for a day to test if that would immediately cause freezing, but it didn't. I have "Use GPU always" enabled in the activity menu and I'm currently only working on the Mapping Cancer Markers project. I don't know if there's a way to select which projects use the GPU. I've suspended GPU usage from the activity menu and web settings for now to see if that changes anything. The default LOWEST priority, is so boinc task yields to anything and everything except when storage i/o is offhand ed and ongoing. If very large checkpoint files are written, you'd notice. Setting to high priority often does not last beyond the next checkpoint, when there's a worker and a control app split such as with ADVina based tasks, otherwise that Hi won't last beyond the end of a task. For sure, after hundreds of millions of MCM1 tasks, it can be said this is one of the most docile, least invasive sciences there are at WCG, short for: whatever is causing this, it's something with your system / software mix, possibly a hyperactive AV. Recommend exclusion of scanning the BOINC data dir (safe, as the data dir is sandboxed) |
||
|
Datanalle
Cruncher Joined: Oct 9, 2015 Post Count: 3 Status: Offline |
1st of all, tell us more about your equipment. Is it a laptop or a PC? What configuration do you run? Desktop PC with: Intel i5-6600K 3.5 GHz, 16 GB RAM, GeForce GTX 1060 DUAL 6GB & 3 case fans. 2nd, you should also run WCG temperature regulated. For example, I useTthrottle , but some others might have alternative options. ![]() The temperatures in my system are well within the operating temperature limits even while gaming. I would check for a rogue process using task manager or process lasso. Check to see if some process is accruing time when it should not be. A long time ago I had a svchost.exe which was accruing time which I think was caused by a botched update, but it was so long ago I can't recall the details. Another possibility is your antivirus maybe checking too often or too aggressively. Cheers I've been keeping an eye on this and there's nothing unusual. System Idle Process always has the most CPU time as expected, next up is usually Firefox, and then any WCG processes with both taking an hour or two of processor time over a day. Other processes beyond these use mere minutes of CPU time. Recommend exclusion of scanning the BOINC data dir (safe, as the data dir is sandboxed) Thanks for pointing that out but there's no change after adding the directory to exclusions. For sure, after hundreds of millions of MCM1 tasks, it can be said this is one of the most docile, least invasive sciences there are at WCG, short for: whatever is causing this, it's something with your system / software mix, possibly a hyperactive AV. That would appear to be the case since I haven't found any other people with this same problem. It's also weird that when these freezes happen none of the CPU cores are necessarily even topped out. There's always a spike in CPU usage but it is not always at 100% usage when freezing. |
||
|
SekeRob
Master Cruncher Joined: Jan 7, 2013 Post Count: 2741 Status: Offline |
From re-reading your OP, you do understand that the BOINC throttle is really a monster designed to lower temps [aimed at then laptops], not to lower load persé. At your 20% it means it will pause 4 whole seconds, then run 1 whole second.
Once upon a time, WCG techs convinced the Berkeley developers to set the control by one tenth of a second fractions, so a 50% of CPU time would render 0.5 seconds pause, run 0.5 seconds, which gave a much smoother experience, BUT, then the developers reversed it, because everything else including the science apps only understood whole second intervals (The whole BOINC world would have to retool), so for a long time ran ThreadMasterGUI service which allowed control at the ms level, plus you could set it at science app level, so light ones could be given bigger slices and heavy/hotter running, smaller time slices. Similar with Process Lasso you could do very precise controlling including affinity. Sadly TMG no longer works on Windows 10. About PL I do not know. |
||
|
KerSamson
Master Cruncher Switzerland Joined: Jan 29, 2007 Post Count: 1679 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Are you sure that the troubles are caused by Boinc?
----------------------------------------I do not operate boinc on my business laptop and since a couple of months I experienced some unexpected freezing of Windows 7 Pro SP1 x64. I investigated the situation and I can only focus on Windows itself or considering that my old CPU is a little bit overwhelmed by current applications, excepted that everything is running fine a large part of the time. Again, maybe you face to a windows problem and not to a boinc-related problem. Cheers, Yves |
||
|
|
![]() |