My workstation has a 6 core cpu with 16GB RAM, I had Charity Engine running in the background but even though it was set to use 50% cpu it was causing spikes that interfered with anything I did. Eventually it seemed like my computer gave up and rebooted and wouldn't boot into Windows 7. So after fixing it (a few hours) got it working then it just died. I had to re-install my OS and I have installed everything bar Charity Engine and I can only attribute the problem with that.
There appears to be no way of making Charity Engine have affinity to just 2 cpus for example. So I now have it running on an old copy of WinXP in a virtual machine which I can then assign it to have affinity to 2 cpu's leaving my workstation usable.
Just a tip for others that may have problems running software in the foreground.
intresting problem if you do reinstall on your computer try going into advanced mode then
tools then into computer prefrences then set "on multiproccesor systems us at most" to 33% this will lock it to 2 cores on your system (100% divided by 6 = 16 so 16% is 1 core) and drop down the CPU time to 95% this will stop any major problems with anything heavy.... just a thought if you want to try it or anyone else has problems
option 2 download the newest versions of BOINC (the software charity engine uses) and attach the charity engine as the account manager this way is the way i do it
also please note not alot of people have noticed this for some reason but with more powerful systems the CPU can out pace the Harddrives on speed so the harddrives can become a bottle neck, first noticed this in windows 7 64bit with a dual I7 server
Sorry to hear about your trouble, interesting work-around with the VM.
Sounds like something else interfering, though. A PC of that power should be able to run Charity Engine - especially at only 50% - without so much as pausing for breath. The spikes in CPU activity are quite normal (if that's what you refer to?).
You can set CE to only use a certain number of cores if you wish, but it's usually more efficient to let the OS decide that for itself, eg. on a quad-core system at 50% CPU, it prefers to use all four cores half of the time, rather than two cores at 100% and two at zero.
(It's always the lowest priority task on the PC, too. Everything else gets CPU time first, CE only gets the leftovers - if anything at all.)
...and thanks to vm1990 for the heads-up about the HDD bottleneck.
We don't recommend changing CPU setting to 95%, by the way. CE default setting is 60%, the efficiency sweet spot as recommended by IBM World Community Grid.
The difference between a CPU running at 5% or 60% usage is only <10% more power, whereas going from 60% to 100% usage needs almost double the power.
That last 40% of CPU is just not worth it. The first 60% - that's the bargain.
My issue came from the engine spiking from 50% to 100% across all cores, on stopping CE I had no problems disk activity was fine (x2 SSD's in RAID0)
In the virtual machine i can have the 2 cores, or even 4 cores dependant on my usage through the day, set as core 5 and 6 and running at 100% leaving the other cores running at near zero load unless I am using them.
My work around was a quick dirty fix, would using your solutions enable me such control?
the solun i suggested wont let you lock tasks to certan cores bt will allow you to control how much of the resorces CE uses if your unsure cheat a bit and test out the solutions inside the VM :) see which one you like then if you want try it on your main computer also i dont know if youve checked but make sure your computers not getting to hot, if the CPU gets to hot it will throttle back and cause seriouse lag on the system
@mark while CE itself uses small amounts of CPU other projects (eistein@home) are allowed to use as much or the free computer resorces as they like so limiting everything to 95% will still let CE use its small portion but will stop the power hog prjects from eating away mass resorces, for some reson when you run tasks in BOINC it takes a few secounds for BOINC to ease off to allow other programs to get there CPU time, its really confusing its like BOINC manager has lower prority so takes longer to notice a program wants to use some extra CPU... it only affects a small amount of systems though and iv only seen it affect high power systems. but droping 1 or even both of the feilds down to 95% normaly allows the system to regain enough control to quickly reallocat CPU time if another program spikes and needs the CPU time..
Thanks vm - I gather you must have an existing BOINC install. I've asked the tech team to contact you, that's a very interesting point you make there.
Peter - did you also have BOINC already? Or is CE the first and only BOINC client on that PC?
To find the settings you want in CE: open the CE app, click 'advanced view', then the 'advanced' tab, then 'preferences' from the drop-down menu. You can tweak everything there: CPU usage, number of cores, when it's allowed to connect, etc.
We actually had a issue with Einstein on some machines, so we dropped it temporarily. It was opening too many images and hogging RAM. (Check your Windows Task Manager, that might be it?)
It's generally not a good idea to run BOINC at 100% constantly for several reasons, only one of them being efficiency. Even at a gentle 60% usage, error rates are still 2%-3% across the grid - often because of overclocked CPUs - and that error rate only increases as CPUs get hotter. A discarded work unit is no good to anyone.
Yes, all PCs should be able to run at 100% for their entire working life, but in practice that's not the case (especially with laptops, now the most common form of PC). The CPU gets hot or overheats, fans kick up to maximum, dust buildup increases - and as you note, the PC can grumble about resource allocation.
The CPU can spike to 100%, that's fine. It heats up and cools down almost instantly, so (for example) 50% usage can be either be constant or pulsing on-off, on-off, on-off - the result is the same: a cooler CPU and no maxed-out fans.
And, as mentioned, running at 60% max uses negligible extra power compared to running at 100%, so it's literally pennies per day to run. The idea is to make CE as unobtrusive as possible; no noticeable effect on the PC or your power bills - unless you want to override the settings, of course.
Nice temps also just asking is it a 4core that youve unlcoked to 6 core or is it a genuine 6 core?? just asking as sometimes in the unlocked 4-6 cores a dogie core can show up when using BOINC
also just asking im assuming with it being 6 core its AMD and some older AMDS like the phonom II do play around a bit when there pushed
and ontop of everything if you do try it back on your main computer and do have problems try disabling and cool&quiet in the bios, iv had problems with them before as they mess up the clock speeds to keep the system cool and it annoys the BOINC engine
@MARK
yes i have a BOINC installed as iv been using it since way before CE was even an idea, my first project was one of the founders of this idea World Community Grid
been with them since late 2008 and a few others
shame about einstein project but like alot of other projects they want get as much done as possible like WCG they use 100% unless you tell it otherwise as for it being a RAM hog iv noticed a few work units using around 200MB RAM but its not that bad unless your either using a pretty old computer or went cheap on the amount of memory theses days i work it out at 1GB per core on a computer as recommended or 512MB of ram per core minimum anything less and you need to upgrade, if you have a DDR3 Board then its shame on you 2GB sticks cost around £11
also @mark "but if it ain't broke, as they say..." .....its going to break soon... Right??? :D
might be a bit of a silly question have you tried stress testing your computer and trying to do somethings while the stress test is going on? might be worth checking out use something like Prime 95 at the very least its going to tell you if theres a bottle neck on your computer or if it is CE
The Prime 95 test will provide a more sustained stress test of the cpu more than most games. Though the instability could be caused by other programs not coded well and are not handling the shared environment.
@peter I'm running the 1090T 6 core AMD and 16Gb ram in my desktop and have only experienced instabilitiy or sluggishness with applications running on the GPU... which is not the case for you as using Charity Engine... which GPU computing is not supported by Charity Engine at this time. Thus that is not likely the issue. I did find for a time Chrome caused some instability and sluggishness as they were developing GPU acceleration but that has come to pass as it matured and I updated my graphics drivers.
Have you checked for driver updates for your systems components? Some of the performance issues could also be a result of a rare driver bug only experienced in certain conditions now being shown by Charity Engine's resource use. Most likely a video driver, BIOS, or chipset would be my guess.
hmmm... I am starting to wonder if something is on its way out... my projects drive (x2 500GB in RAID 1) just went critical and 1 drive has become corrupted.
Memory all tests fine, cpu isn't over heating, PSU safety cut out twice this evening, both RAID 1 HDs show as Healthy, gfx card isn't over heating, SSDs are fine as are my x2 storage drives... motherboard could be on its way out.
Could be I am just trying to put too much through it and it started to say enough is enough, best replace it before it take out any other components. :)
new motherboard, all other parts tested over the last 1.5 months with zero problems, install charity engine... instant death, can't do anything on the machine as it just crashes it shortly after boot.
Stopped CE from starting at boot, everything is stable until I start CE and it crashes the machine... so its uninstalled permanently.
Thank you for keeping us informed of your progress. If you would send a complete list of hardware and software of your machine's configuration including brands, models, versions etc we would be able to look into what might be causing the conflict as your testing has ruled out the hardware to the best of our knowledge.
You can sent the information via our Contact Us to protect your computer's information.
This slowdown/crash is very strange behavior that has not appeared before for any other user and we would like to look for what might be causing this behavior. There is likely a piece of software that is not behaving properly when running along side Charity Engine or a driver issue only appearing under load or certain calculations.
Just got my computer back up running, having spent my weekend re-installing. I disabled CE to see if it made my machine stable, which it appeared to, I then uninstalled it and shut down Friday night only to find it wouldn't get passed windows log in and wouldn't go in to safe mode.
The only thing that I can see making a difference between me running it on XP virtualized and on this machine is that I don't have a page file... which I realized I forgot to re-enable after installing everything previously.
My workstation has a 6 core cpu with 16GB RAM, I had Charity Engine running in the background but even though it was set to use 50% cpu it was causing spikes that interfered with anything I did. Eventually it seemed like my computer gave up and rebooted and wouldn't boot into Windows 7. So after fixing it (a few hours) got it working then it just died. I had to re-install my OS and I have installed everything bar Charity Engine and I can only attribute the problem with that.
There appears to be no way of making Charity Engine have affinity to just 2 cpus for example. So I now have it running on an old copy of WinXP in a virtual machine which I can then assign it to have affinity to 2 cpu's leaving my workstation usable.
Just a tip for others that may have problems running software in the foreground.
intresting problem if you do reinstall on your computer try going into advanced mode then
tools then into computer prefrences then set "on multiproccesor systems us at most" to 33% this will lock it to 2 cores on your system (100% divided by 6 = 16 so 16% is 1 core) and drop down the CPU time to 95% this will stop any major problems with anything heavy.... just a thought if you want to try it or anyone else has problems
option 2 download the newest versions of BOINC (the software charity engine uses) and attach the charity engine as the account manager this way is the way i do it
also please note not alot of people have noticed this for some reason but with more powerful systems the CPU can out pace the Harddrives on speed so the harddrives can become a bottle neck, first noticed this in windows 7 64bit with a dual I7 server
hope something out of all thats useful :)
Hi Peter,
Sorry to hear about your trouble, interesting work-around with the VM.
Sounds like something else interfering, though. A PC of that power should be able to run Charity Engine - especially at only 50% - without so much as pausing for breath. The spikes in CPU activity are quite normal (if that's what you refer to?).
You can set CE to only use a certain number of cores if you wish, but it's usually more efficient to let the OS decide that for itself, eg. on a quad-core system at 50% CPU, it prefers to use all four cores half of the time, rather than two cores at 100% and two at zero.
(It's always the lowest priority task on the PC, too. Everything else gets CPU time first, CE only gets the leftovers - if anything at all.)
...and thanks to vm1990 for the heads-up about the HDD bottleneck.
We don't recommend changing CPU setting to 95%, by the way. CE default setting is 60%, the efficiency sweet spot as recommended by IBM World Community Grid.
The difference between a CPU running at 5% or 60% usage is only <10% more power, whereas going from 60% to 100% usage needs almost double the power.
That last 40% of CPU is just not worth it. The first 60% - that's the bargain.
My issue came from the engine spiking from 50% to 100% across all cores, on stopping CE I had no problems disk activity was fine (x2 SSD's in RAID0)
In the virtual machine i can have the 2 cores, or even 4 cores dependant on my usage through the day, set as core 5 and 6 and running at 100% leaving the other cores running at near zero load unless I am using them.
My work around was a quick dirty fix, would using your solutions enable me such control?
the solun i suggested wont let you lock tasks to certan cores bt will allow you to control how much of the resorces CE uses if your unsure cheat a bit and test out the solutions inside the VM :) see which one you like then if you want try it on your main computer also i dont know if youve checked but make sure your computers not getting to hot, if the CPU gets to hot it will throttle back and cause seriouse lag on the system
@mark while CE itself uses small amounts of CPU other projects (eistein@home) are allowed to use as much or the free computer resorces as they like so limiting everything to 95% will still let CE use its small portion but will stop the power hog prjects from eating away mass resorces, for some reson when you run tasks in BOINC it takes a few secounds for BOINC to ease off to allow other programs to get there CPU time, its really confusing its like BOINC manager has lower prority so takes longer to notice a program wants to use some extra CPU... it only affects a small amount of systems though and iv only seen it affect high power systems. but droping 1 or even both of the feilds down to 95% normaly allows the system to regain enough control to quickly reallocat CPU time if another program spikes and needs the CPU time..
Thanks vm - I gather you must have an existing BOINC install. I've asked the tech team to contact you, that's a very interesting point you make there.
Peter - did you also have BOINC already? Or is CE the first and only BOINC client on that PC?
To find the settings you want in CE: open the CE app, click 'advanced view', then the 'advanced' tab, then 'preferences' from the drop-down menu. You can tweak everything there: CPU usage, number of cores, when it's allowed to connect, etc.
We actually had a issue with Einstein on some machines, so we dropped it temporarily. It was opening too many images and hogging RAM. (Check your Windows Task Manager, that might be it?)
It's generally not a good idea to run BOINC at 100% constantly for several reasons, only one of them being efficiency. Even at a gentle 60% usage, error rates are still 2%-3% across the grid - often because of overclocked CPUs - and that error rate only increases as CPUs get hotter. A discarded work unit is no good to anyone.
Yes, all PCs should be able to run at 100% for their entire working life, but in practice that's not the case (especially with laptops, now the most common form of PC). The CPU gets hot or overheats, fans kick up to maximum, dust buildup increases - and as you note, the PC can grumble about resource allocation.
The CPU can spike to 100%, that's fine. It heats up and cools down almost instantly, so (for example) 50% usage can be either be constant or pulsing on-off, on-off, on-off - the result is the same: a cooler CPU and no maxed-out fans.
And, as mentioned, running at 60% max uses negligible extra power compared to running at 100%, so it's literally pennies per day to run. The idea is to make CE as unobtrusive as possible; no noticeable effect on the PC or your power bills - unless you want to override the settings, of course.
Mark - CE is the first and only, I downloaded a new file and installed recently so I guess that means it most up to date.
CPU runs at a cool 41 on air under full load on all cores and the case internals get a clean every few months.
I think I'll keep it running in a VM as if causes instability it should only affect the VM.
41 with six fully-loaded cores? Wow. That's one effective cooler you have there. Nice to know some PCs are still being built properly...!
I guess you've got it working, so no need to tweak it further.
You could always set CPU and cores both to zero, then bring them online one at a time - but if it ain't broke, as they say...
Nice temps also just asking is it a 4core that youve unlcoked to 6 core or is it a genuine 6 core?? just asking as sometimes in the unlocked 4-6 cores a dogie core can show up when using BOINC
also just asking im assuming with it being 6 core its AMD and some older AMDS like the phonom II do play around a bit when there pushed
and ontop of everything if you do try it back on your main computer and do have problems try disabling and cool&quiet in the bios, iv had problems with them before as they mess up the clock speeds to keep the system cool and it annoys the BOINC engine
@MARK
yes i have a BOINC installed as iv been using it since way before CE was even an idea, my first project was one of the founders of this idea World Community Grid
been with them since late 2008 and a few others
shame about einstein project but like alot of other projects they want get as much done as possible like WCG they use 100% unless you tell it otherwise as for it being a RAM hog iv noticed a few work units using around 200MB RAM but its not that bad unless your either using a pretty old computer or went cheap on the amount of memory theses days i work it out at 1GB per core on a computer as recommended or 512MB of ram per core minimum anything less and you need to upgrade, if you have a DDR3 Board then its shame on you 2GB sticks cost around £11
also @mark "but if it ain't broke, as they say..." .....its going to break soon... Right??? :D
Its a true 6 core, AMD Phenom II 1055T not overclocked, cool and quiet always disabled as I have never trusted it.
oh the cpu runs at 19 degrees idling in Windows considering ambient temp here is 24 its doing its job well :D
might be a bit of a silly question have you tried stress testing your computer and trying to do somethings while the stress test is going on? might be worth checking out use something like Prime 95 at the very least its going to tell you if theres a bottle neck on your computer or if it is CE
Well I haven't run prime but I regularly play games or use the internet and listen to mp3's whilst all cores are rendering and I am waiting.
Now that's multi-tasking... :)
A correction: it was Rosetta that caused RAM issues, not Einstein.
Yes, it was Rosetta@home with the RAM issues.
The Prime 95 test will provide a more sustained stress test of the cpu more than most games. Though the instability could be caused by other programs not coded well and are not handling the shared environment.
@peter I'm running the 1090T 6 core AMD and 16Gb ram in my desktop and have only experienced instabilitiy or sluggishness with applications running on the GPU... which is not the case for you as using Charity Engine... which GPU computing is not supported by Charity Engine at this time. Thus that is not likely the issue. I did find for a time Chrome caused some instability and sluggishness as they were developing GPU acceleration but that has come to pass as it matured and I updated my graphics drivers.
Have you checked for driver updates for your systems components? Some of the performance issues could also be a result of a rare driver bug only experienced in certain conditions now being shown by Charity Engine's resource use. Most likely a video driver, BIOS, or chipset would be my guess.
Ah, so animating in Maya would potentially cause sluggishness with CE running in the background on all 6 cores?
Although that wouldn't explain my ability to render in Maya on 6 cores and play a 3D game at the same time.
hmmm... I am starting to wonder if something is on its way out... my projects drive (x2 500GB in RAID 1) just went critical and 1 drive has become corrupted.
Memory all tests fine, cpu isn't over heating, PSU safety cut out twice this evening, both RAID 1 HDs show as Healthy, gfx card isn't over heating, SSDs are fine as are my x2 storage drives... motherboard could be on its way out.
When you have eliminated the impossible, whatever remains - however improbable - is usually the motherboard...
You have my sympathy. Been there!
Could be I am just trying to put too much through it and it started to say enough is enough, best replace it before it take out any other components. :)
Woo hoo new motherboard! :D
That's the spirit!
Shame we can't change the name of this thread, though... ;)
theres no admin function to this forum? or access to the database for direct editing?
Don't know if that would look worse? Censorship and all that.
Then again, some people will just see the title in the forum. We've got another similar thread just started.
When a PC goes wrong, first suspect is always the most recent software install. (And last one is the motherboard!)
new motherboard, all other parts tested over the last 1.5 months with zero problems, install charity engine... instant death, can't do anything on the machine as it just crashes it shortly after boot.
Stopped CE from starting at boot, everything is stable until I start CE and it crashes the machine... so its uninstalled permanently.
Thank you for keeping us informed of your progress. If you would send a complete list of hardware and software of your machine's configuration including brands, models, versions etc we would be able to look into what might be causing the conflict as your testing has ruled out the hardware to the best of our knowledge.
You can sent the information via our Contact Us to protect your computer's information.
This slowdown/crash is very strange behavior that has not appeared before for any other user and we would like to look for what might be causing this behavior. There is likely a piece of software that is not behaving properly when running along side Charity Engine or a driver issue only appearing under load or certain calculations.
Just got my computer back up running, having spent my weekend re-installing. I disabled CE to see if it made my machine stable, which it appeared to, I then uninstalled it and shut down Friday night only to find it wouldn't get passed windows log in and wouldn't go in to safe mode.
The only thing that I can see making a difference between me running it on XP virtualized and on this machine is that I don't have a page file... which I realized I forgot to re-enable after installing everything previously.
Is not haivng a paging file the like cause?
?