Stuck at 100%

Having problems installing that new stick of memory? Found some great software or having issues with something? Or maybe want to chat about your PlayStation, X-Box, Nintendo, Sega, even your old Spectrum 48k....! Or maybe something you want to sell or acquire (computing related of course!). Let us know here...
Post Reply
rich2000k
UBT Contributor
Posts: 107
Joined: Sun Dec 15, 2013 12:00 am
Location: Staffordshire

Stuck at 100%

Post by rich2000k »

All my CPU tasks are getting to 99% (in the estimated time) then take hours and hours to go to 100% on Milkyway and now started the same on BCU CPU/GPU crossover tasks. When they eventually get to 100% they never "finish" and report. They can be literally sitting there for 19hrs+, stuck at 100%...can anyone help?, have I got something set wrong.

I have now changed milkyway to just GPU only, but this is really annoying. I was getting 600,000+ credit per day and now down in the 10's of thousands.

Luckily I have other projects now picking up the spare CPU time.
UBT - Timbo
UBT Forum Admin
Posts: 9706
Joined: Mon Mar 13, 2006 12:00 am
Location: NW Midlands
Contact:

Post by UBT - Timbo »

Hi Rich,

I had exactly the same problem with SRBase.

http://srbase.myfirewall.org/sr5/forum_ ... php?id=326

It *seems* that it might be down to "resources", either lack of available RAM, or disk space, or something else, such as browser "tabs" hanging onto memory.

One option is to try exiting BOINC Manager, rebooting, and see if that helps.

What the specification of the PC that has the problem ?

regards
Tim
UBT - Mikee
Marvin the Dalek
Posts: 4399
Joined: Wed Mar 15, 2006 12:00 am
Location: North Wales

Post by UBT - Mikee »

I get that a few times a week with BitCoin. If I see it's stuck I just abort it, otherwise it'll hang round at 100% for up to 8 hours.

I've looked at the results of these and some of them had a 'called boinc_finish' after just a few minutes but they carry on like a Duracell bunny!

Mike
Follow us on Twitter... http://twitter.com/UKBOINCTeam

Image Image

Image
rich2000k
UBT Contributor
Posts: 107
Joined: Sun Dec 15, 2013 12:00 am
Location: Staffordshire

Post by rich2000k »

My system is pretty good, it's as follows:
AMD 8350 cpu
Sapphire R9 290 gpu
1.2Tb HDD's split over 2 drives, 1TB Partitioned into 2 500gb.
8GB DDR5 Ram
XFX 850 Gold psu

I'd be surprised if it's the system causing it but I can have a look at increasing the ram and memory. Boinc is allowed to have. I deal with hard drives in my spare time so have the odd 500GB drive lying around. I may just plug one in and let Boinc have it all to itself.

Thanks for the info i'll keep you posted.

Rich
rich2000k
UBT Contributor
Posts: 107
Joined: Sun Dec 15, 2013 12:00 am
Location: Staffordshire

Post by rich2000k »

Rebooting boinc doesn't work, it just goes back to 0% and starts again...humm this one has me stumpted.
UBT - Timbo
UBT Forum Admin
Posts: 9706
Joined: Mon Mar 13, 2006 12:00 am
Location: NW Midlands
Contact:

Post by UBT - Timbo »

rich2000k wrote:My system is pretty good, it's as follows:
AMD 8350 cpu
Sapphire R9 290 gpu
1.2Tb HDD's split over 2 drives, 1TB Partitioned into 2 500gb.
8GB DDR5 Ram
XFX 850 Gold psu

I'd be surprised if it's the system causing it but I can have a look at increasing the ram and memory. Boinc is allowed to have. I deal with hard drives in my spare time so have the odd 500GB drive lying around. I may just plug one in and let Boinc have it all to itself.

Thanks for the info i'll keep you posted.

Rich
Hi Rich,

Thanks for the info.

So, the AMD CPU is an 8-core.

If you are using your GPU as well for BOINC, then the max number of CPU's available (for CPU tasks) is 7. If you are doing any other work on your PC, then it could even be less

Have you set up a app_info.xml file for each CPU project, in order to limit the number of CPU cores to 7? Or otherwise, have you set the BOINC Manager "Preferences" to use upto 87.5% (ie 7 cores) of your CPU resources.

If neither of the above applies, then it *could* be that when the task finished, there's no CPU "grunt" available to write the file to disk, before it submits it to the respective project server.

Otherwise, with 8Gb of RAM and plenty of free HDD space, I can't see there is a big problem. Reb on the SRBase forum, thought I might have too much RAM allocated to Firefox/Chrome and this was causing the issue I was seeing. Having done a reboot plus making sure FF/Chrome were not used, I've not seen the problem again.

Hope this helps

regards
Tim
rich2000k
UBT Contributor
Posts: 107
Joined: Sun Dec 15, 2013 12:00 am
Location: Staffordshire

Post by rich2000k »

Cheers Tim

You are correct 8 cores (until I win the lottery, then who knows how much I'll get lol)

I've gone through and set them all at 85% see if it helps at all.

I'll keep you posted, but for the moment it's behaving
Saying that it's been doing nothing else all day.
rich2000k
UBT Contributor
Posts: 107
Joined: Sun Dec 15, 2013 12:00 am
Location: Staffordshire

Post by rich2000k »

Right that didn't work

This morning a BCU cpu/gpu wu used 1 sceond of cpu but was still running 43 mins later again at 100%.....ooohh this is driving me potty!
UBT - Timbo
UBT Forum Admin
Posts: 9706
Joined: Mon Mar 13, 2006 12:00 am
Location: NW Midlands
Contact:

Post by UBT - Timbo »

rich2000k wrote:Right that didn't work

This morning a BCU cpu/gpu wu used 1 sceond of cpu but was still running 43 mins later again at 100%.....ooohh this is driving me potty!
Hi Rich,

:Looking at your recent BU results here:

http://www.bitcoinutopia.net/bitcoinuto ... e=4&appid=

They are mostly all finishing OK....so, it may just be the "odd one" that has an issue - this can be down to all sorts of thigns: BU server is busy, internet goes down briefly, ipominer.com website offline or overloaded, etc etc.

I think that for the short-term, just keep an eye on it and given the approx crunching times of your open_cl tasks (taking between 1-2 minutes, with one taking about 10 mins) if you see any WU's that are stuck after (say) 15 mins, then just suspend them and leave them be for a while....

Obviously, one loses the actual crunched time to date, but OTOH, if it's just sitting there doing nothing then better for you to be running a different task in the meantime.

regards
Tim
Last edited by UBT - Timbo on Fri May 29, 2015 9:17 pm, edited 1 time in total.
rich2000k
UBT Contributor
Posts: 107
Joined: Sun Dec 15, 2013 12:00 am
Location: Staffordshire

Post by rich2000k »

YAY!, We're all working again.
UBT - Timbo
UBT Forum Admin
Posts: 9706
Joined: Mon Mar 13, 2006 12:00 am
Location: NW Midlands
Contact:

Post by UBT - Timbo »

rich2000k wrote:YAY!, We're all working again.
Great news....with luck, these issues were just a blip.

I also noticed that yesterday, BU (or mostly likely one of the pools) went down.

Sadly, I was away for most of the day and night, and my PC couldn't restart any BU tasks, though it should have restarted automatically....not sure why it didn't.

But despite losing nearly a days worth of BU credits, I'm back now.

regards
Tim
rich2000k
UBT Contributor
Posts: 107
Joined: Sun Dec 15, 2013 12:00 am
Location: Staffordshire

Post by rich2000k »

I did notice that it was 90% of the time milkyway that was stuck...so bad to the point I re-did the same single task 5 times (every-time I rebooted or even paused boinc it restarted the task and got to 100% then froze).

Easiest answer was to abort the project and run 1 that works.

Hence why i'm motoring again.
UBT - Timbo
UBT Forum Admin
Posts: 9706
Joined: Mon Mar 13, 2006 12:00 am
Location: NW Midlands
Contact:

Post by UBT - Timbo »

Hi Rich,

There can be all sorts of reasons why WU's fail to compute correctly...whether it's an issue with the downloaded task, some memory/hardware issues, or even trouble uploading the result.

The main thing is to keep an eye on it and if you need to abort a WU, then do so...there's enough computing power around to ensure that that WU will be crunched by another member, in due course.

If everything is tickety-boo now, then good news :)

regards
Tim
rich2000k
UBT Contributor
Posts: 107
Joined: Sun Dec 15, 2013 12:00 am
Location: Staffordshire

Post by rich2000k »

this is really bugging me now....it's so slow now my gpu isn't even kicking it's fans in as it's not being utilized enough. If I even snooze Boinc for just a sec it cancels the task and restarts it....something is amiss here.

Last night I restarted my PC as did the new windows update and immediately it fired up crunching at full pelt, then again this morning next to nothing and issues again.

Come on guys help me out here something is either wrong with my PC or setting and I can't figure out what.

CPU tasks btw are again fine just GPU having issues.
I have checked my RAM usage and never had less than 5gb free at any one time, space on HDD for Boinc still has 38gb free.

I can't warrant leaving my PC on all day for just cpu crunching it has to be worth it.
UBT - Timbo
UBT Forum Admin
Posts: 9706
Joined: Mon Mar 13, 2006 12:00 am
Location: NW Midlands
Contact:

Post by UBT - Timbo »

Hi Rich,

OK - I can only guess at what might be the issue, if it's the GPU that's now causing issues.

1) You could try a different project that has GPU tasks available...such as Collatz, Asteroids or SETI. Just make sure you go to the respective projects website and then "allow" the ATI/AMD GPU applications. Likewise, you could switch MW apps to their "Separation" tasks...these don't take long at all (less than 2 mins) and you get lots of lovely credits.

You only need do for say a few hours, just to make sure that your GPU set-up is OK.

2) I used to run ATI (now AMD) cards, but found their software (Catalyst) could sometimes be an issue (though not with BOINC I must say). Given you have a  damn fine "high-end" card, (and assuming it's not gone faulty), then the obvious thing to do is to upgrade to the latest GPU drivers....or perhaps even to "downgrade" to a previous (known working) version.

3) One issue I found with the NVidia cards I now use, is that they can put out copious amounts of heat, which given that it's been quite warm recently, means that sometimes the cards can struggle a bit.

I've now installed the MSI "Afterburner" app which gives one more precise control of GPU voltages and frequencies...in fact I use this to "under-run" the GPU's in order to minimise their heat output.

So, there might be an alternative to Afterburner that can control the AMD/ATI cards.

EDIT: In fact I believe MSI Afterburner will control AMD cards too...

http://gaming.msi.com/features/afterburner

I'm sure some others might also be able to suggest other options, so I'll leave you with the above for now.

regards
Tim
rich2000k
UBT Contributor
Posts: 107
Joined: Sun Dec 15, 2013 12:00 am
Location: Staffordshire

Post by rich2000k »

Right I gave that a try with Seti and a 21 min task is taking 1:31.42....so my GPU is not running at full whack or their eta is WAYYYY off...humm

Right best option is i'll download a high intensity game for my system and test the GPU on that for a day or so (no crunching for 48hrs). That'll give me some idea as to if it is the GPU for sure.
UBT - Timbo
UBT Forum Admin
Posts: 9706
Joined: Mon Mar 13, 2006 12:00 am
Location: NW Midlands
Contact:

Post by UBT - Timbo »

Hi Rich,

If the GPU is giving you a long "time to completion" it could be an idea to check to see if the GPU is being throttled back, due to (say) a fan failure...maybe also check the PCI-e connectors and see if the PSU is actually powering the graphics card with the correct voltage.

Also: If you try the Afterburner software, it will actually show you the %age rate at which the GPU and CPU's are running....so, you can keep the basic window open (say the right side of your screen) and then start and stop various BOINC tasks and see what effect it has on the visible graphs - it's actually very useful to see a graphical display of what's going on.

Of course, if you are running Win7, then the extended section of Task Manager also shows this too.

regards
Tim
Post Reply