| Author |
Message |
|---|
|
|
|
I'm currently running some CPU Genefer WU as part of the challenge.
I have an i7 that is running BOINC 6.12.34 on Windows 2008 Server that is set to NOT allow computing while the machine is in use.
The PG-Genefer WU's do not pause when start using the machine, BOINC thinks they are paused, but a check with Task Manager shows that the 6 WUs are still using CPU.
The GeneferCUDA WU's, however do pause.
WU from other projects that have been running do stop using the CPU when the machine is in use.
Stopping the BOINC client does cause the running processes to go away, but I have seen the CPU pegged because BOINC restarted some other projects instead of the PG Genefer units.
I'm planning to upgrade the BOINC client to the v7 stream but just thot folks would want to know about this issue.
____________
|
|
|
|
|
|
Upgrading to BOINC 7.0.25 won't solve your problem. GFN CPU is the only project I think I've ever seen this behavior.
____________
PrimeGrid Challenge Overall standings --- Last update: From Pi to Paddy (2016)
 |
|
|
 pschoefer Volunteer developer
Volunteer tester
Send message
Joined: 20 Sep 05
Posts: 686
ID: 845
Credit: 3,023,173,815
RAC: 1,300,805
                              
|
|
For me it does suspend properly, if I uncheck BOINC's "Leave apps in memory" option.
____________
  |
|
|
|
|
Upgrading to BOINC 7.0.25 won't solve your problem. GFN CPU is the only project I think I've ever seen this behavior.
I think that might be caused by Boinc's "panic mode". Usually, Boinc thinks that genefers will take hundreds of hours to complete, way beyond the deadline, so it keeps crunching them even if suspended, unless boinc client is restarted. I've seen that happen with other wu's, from different projects (Einstein@home, I think).
____________
676754^262144+1 is prime |
|
|
|
|
|
Hmm, that's a daft feature. Guestimated time left is totally within the deadline. 1 to 2 days left on a 6th of July deadline. So it shouldn't be going crazy over it, now should it?
____________
PrimeGrid Challenge Overall standings --- Last update: From Pi to Paddy (2016)
|
|
|
|
|
Upgrading to BOINC 7.0.25 won't solve your problem. GFN CPU is the only project I think I've ever seen this behavior.
I think that might be caused by Boinc's "panic mode". Usually, Boinc thinks that genefers will take hundreds of hours to complete, way beyond the deadline, so it keeps crunching them even if suspended, unless boinc client is restarted. I've seen that happen with other wu's, from different projects (Einstein@home, I think).
No, it's not panic or high priority mode, since BOINC believes the WU is suspended due to computer use. It's just that the WU doesn't suspend when told to do so.
____________
|
|
|
|
|
|
Possibly it could be caused by the 'normal' process priority in stead of the low priority.
I'm not running a CPU task, but the Genefer GPU-task is running with the 'normal' priority, which is atypical for BOINC-processes. |
|
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 14045
ID: 53948
Credit: 482,681,090
RAC: 578,366
                       
|
Possibly it could be caused by the 'normal' process priority in stead of the low priority.
I'm not running a CPU task, but the Genefer GPU-task is running with the 'normal' priority, which is atypical for BOINC-processes.
Are you certain Genefx64 is running at normal priority? It shouldn't do that; it should be at idle priority like all other CPU apps. Only GPU apps should be at normal priority (so they CPU can keep the GPU fed.)
If so, what OS are you using and what version of BOINC? And if you're using app_info, what version of Genefer are you running?
____________
My lucky number is 75898524288+1 |
|
|
kiskaVolunteer tester
Send message
Joined: 13 Apr 12
Posts: 47
ID: 138397
Credit: 244,110,791
RAC: 194,215
                 
|
|
Actually whatever dist. your are using whether it is Linux, Mac or Windows BOINC tasks are on a very low priority. In Linux the very low priority is nice: 19 on windows it depends on what task manager you use whether it is the default for 3rd party it depends, I use cmd for task manager so it code is -19 usually on Win systems on Mac I don't know. Hope that solves the priorities. As for it not suspending that is because it is kept in memory and the cpu thinks it still needs crunching where it is on suspend. The gpu tasks suspend due to the fact once the mouse starts moving or something similar action the gpu is directed at rendering your desktop and not on CUDA tasks/workunits. From the logs he is using version 1.07 |
|
|
|
|
Are you certain Genefx64 is running at normal priority? It shouldn't do that; it should be at idle priority like all other CPU apps. Only GPU apps should be at normal priority (so they CPU can keep the GPU fed.)
If so, what OS are you using and what version of BOINC? And if you're using app_info, what version of Genefer are you running?
As I said: I'm not running Genefer CPU's, but only GPU.
So the 'normal' priority for a GPU-task is OK.
For other BOINC-projects I had to set the CPU-process to feed the GPU to normal by 'hand' (eFMer's Priority)
____________
 |
|
|
|
|
|
I have the same thing happening. Both my AMD 1100Ts and my AMD 970 with all 3 on different motherboards and memory. All 3 have BOINC 7.0.25 and Win 7 x64 bit OS. I've tried different priorities but it makes no difference.
Instead of restarting BOINC I go into Win 7s Resource Monitor and suspend each manually but you have to know the PID number. Resuming from it works fine as well (got to remember to resume in BOINC too).
Luckily I'm only running the CPU version for this challenge or this would be more than just a nuisance. The BOINC devs need to fix this, especially since it's happening on other projects as well as mentioned above by Usucapio Libertatis.
NM*
____________
Largest Primes to Date:
As Double Checker: SR5 109208*5^1816285+1 Dgts-1,269,534
As Initial Finder: SR5 243944*5^1258576-1 Dgts-879,713
 |
|
|
|
|
|
Hi computerguy09,
Not sure entirely what is happening here, but can you please upgrade to the latest BOINC client (7.0.x) and try again. If you can still recreate the problem here we'll investigate further.
Cheers
- Iain
____________
Twitter: IainBethune
Proud member of team "Aggie The Pew". Go Aggie!
3073428256125*2^1290000-1 is Prime! |
|
|
|
|
|
Hi, I've the same behavior with Boinc 6.12.33 on Win7 X64 4 Core
PrimGrid Genefer 1.07 is in Status waiting and ProcessExplorer shows that it's running with 100% on a Core.
Two other project results (of 4 with status running) are in status running use only 50% of the CPU time.
The Genefer 1.07 CPU App does not handle the status change of the boinc client correct.
I've seen this behavior already at the challenge.
There are 5 results running on my 4 Core CPU.
Please fix this.
Edit: The Process Priority is Idle (4) as expected
____________
Matthias
 |
|
|
kiskaVolunteer tester
Send message
Joined: 13 Apr 12
Posts: 47
ID: 138397
Credit: 244,110,791
RAC: 194,215
|
|
Well the best I can say is lower the priority to 1 or 2 and see what happens. If nothing happens terminate the process. I did it to Linux and it didn't affect it but I don't know.
Also in the setting is it checked to 'leave apps in memory after suspension' if it is then uncheck and in the 'Activity' menu use the option 'Run based on preferences.' that should stop the app from running.
If not then terminate one from Task Manager but at a rate of: one at a time. I don't know what will happen in Windows since I am Linux. But as I said it shouldn't cause a compute error. BUT just try one at a time terminate/end and see the result of that; if its causing a compute error then don't continue if it is stopping it then terminate all the remaining then instruct BOINC to 'resume' just as before one at a time. |
|
|
|
|
|
Just to say I haven't forgotten this issue, but some 'real life' is getting between me and building a debug version for you. Hopefully in the next week...
____________
Twitter: IainBethune
Proud member of team "Aggie The Pew". Go Aggie!
3073428256125*2^1290000-1 is Prime! |
|
|
|
|
|
Iain -
You might want to hold off on that - Based on a quick Google search this problem seems to be affecting not only PG, but SETI, Rosetta, Docking and others!!
Search phrase was boinc not pausing suspended work units
____________
There's someone in our head but it's not us. |
|
|
|
|
|
When I supspend Genefer(CPU), if keep running without stop
More infomation can found in the video below:
http://www.youtube.com/watch?v=squTRKMNMdo
(sorry, i can't embed it)
System info:
 |
|
|
|
|
|
Finally got to the bottom of this issue and have made a fix. In the process of creating new builds now and will release them to BOINC soon.
Cheers
- Iain
____________
Twitter: IainBethune
Proud member of team "Aggie The Pew". Go Aggie!
3073428256125*2^1290000-1 is Prime! |
|
|
|
|
|
New builds containing the fix have been released, and I hope they will soon be available via BOINC.
Cheers
- Iain
____________
Twitter: IainBethune
Proud member of team "Aggie The Pew". Go Aggie!
3073428256125*2^1290000-1 is Prime! |
|
|
|
|
|
Were these new builds ever released?
I have just started running Genefer CPU work units as the 321 Blast Off challenge was winding down.
I found that if I suspend the Genefer Work Unit or if the BOINC Manager pauses/suspends the Genefer task to run another task then it just keeps going, not releasing the CPU.
This shares the CPU across more tasks than the number of CPU's which gives less than 100% to each task which means all tasks then slow down with the reduced CPU allocation.
I get it all working properly when I restart BOINC Client. This is annoying and I can't be doing that all the time.
I am using Linux Fedora 16 on a 64 bit build running BOINC Client version 6.10.58.
I mentioned this during the challenge in the 321 Blast Off challenge thread, but got no response.
Thanks
Conan
____________
 |
|
|
|
|
|
Hi, the new builds are still not live - sorry! Unfortunately we haven't been able to add any new apps (Genefer or LLR) for several months now :(
The workaround of unchecking the 'suspended apps remain in memory' box should still work for this problem.
Cheers
- Iain
____________
Twitter: IainBethune
Proud member of team "Aggie The Pew". Go Aggie!
3073428256125*2^1290000-1 is Prime! |
|
|
|
|
Hi, the new builds are still not live - sorry! Unfortunately we haven't been able to add any new apps (Genefer or LLR) for several months now :(
The workaround of unchecking the 'suspended apps remain in memory' box should still work for this problem.
Cheers
- Iain
Thanks Iain,
I will just finish what 321 work I have and do Genefer only till I reach at least a Silver badge, then I wont have the problem till I change Applications or projects.
I will be leaving the "remain in Memory" as it is, due to the other projects that I run.
Conan
____________
|
|
|
