hello all,

Yesterday i saw a strange behavior and i dont know if thats normal or not, and how can i solve that....

When i came back from work, i saw that BOIC was only using the GPU, and no CPU core (my CPU is a Intel Core i3 @ 2.93 GHz, GPU is a nVidia GT 220)

why does BOIC behave in that way? how can i solve it?

many thanks

We can probably help you understand this better if you can give us more information.

1) Which (numbered) version of BOINC do you have installed?
2) Which project(s) are you attached to?
3) Which project(s) have work downloaded and available to run?
4) Which project is running a task on the GPU?

1) Which (numbered) version of BOINC do you have installed?

6.10.18 Windows 32-bits

2) Which project(s) are you attached to?

Seti@home and Docking@Home
3) Which project(s) have work downloaded and available to run?

seti@home

4) Which project is running a task on the GPU?

seti@home

btw, im using Windows 7 Home Premium, for if that info is useful....

I've not heard of Windows 7 causing any problems at SETI: there have been issues at Einstein, but they use an unusual application structure.

One thing to check: do you have a file cc_config.xml in your BOINC Data folder, and if so, does it contain the line

<ncpus>0</ncpus>

in the Options section? If so, change it to

<ncpus>-1</ncpus>

(that's minus one) and restart BOINC.

Otherwise, two more questions:

5) How long ago did you install v6.10.18, and has it ever processed SETI on CPU since then?
6) Could you please post the initial lines from the 'Messages' tab of BOINC after a restart, and any extra lines which show any sign of an error.

hi,
i cant find that cc_config.xml file....

Answers to your questions are:

5) i just bought this new PC few days ago, and started BOINC 2 days ago. Yes, it has been processing WUs for this period of time.

6) here they are:

12/01/2010 23:00:56 Starting BOINC client version 6.10.18 for windows_intelx86
12/01/2010 23:00:56 log flags: file_xfer, sched_ops, task
12/01/2010 23:00:56 Libraries: libcurl/7.19.4 OpenSSL/0.9.8l zlib/1.2.3
12/01/2010 23:00:56 Data directory: C:\ProgramData\BOINC
12/01/2010 23:00:56 Running under account Elhoir
12/01/2010 23:00:56 Processor: 4 GenuineIntel Intel(R) Core(TM) i3 CPU 530 @ 2.93GHz [x86 Family 6 Model 37 Stepping 2]
12/01/2010 23:00:56 Processor: 256.00 KB cache
12/01/2010 23:00:56 Processor features: fpu tsc pae nx sse sse2 pni mmx
12/01/2010 23:00:56 OS: Microsoft Windows 7: Home Premium x86 Edition, (06.01.7600.00)
12/01/2010 23:00:56 Memory: 2.99 GB physical, 5.98 GB virtual
12/01/2010 23:00:56 Disk: 444.66 GB total, 415.62 GB free
12/01/2010 23:00:56 Local time is UTC +1 hours
12/01/2010 23:00:56 NVIDIA GPU 0: GeForce GT 220 (driver version 19107, CUDA version 2030, compute capability 1.2, 1024MB, 128 GFLOPS peak)
12/01/2010 23:00:56 Not using a proxy
12/01/2010 23:00:56 Docking URL http://docking.cis.udel.edu/; Computer ID 55817; resource share 100
12/01/2010 23:00:56 SETI@home URL http://setiathome.berkeley.edu/; Computer ID 5257895; resource share 100
12/01/2010 23:00:56 SETI@home General prefs: from SETI@home (last modified 11-Jan-2010 16:34:53)
12/01/2010 23:00:56 SETI@home Computer location: home
12/01/2010 23:00:56 SETI@home General prefs: no separate prefs for home; using your defaults
12/01/2010 23:00:56 Reading preferences override file
12/01/2010 23:00:56 Preferences limit memory usage when active to 1531.56MB
12/01/2010 23:00:56 Preferences limit memory usage when idle to 2756.80MB
12/01/2010 23:00:56 Preferences limit disk usage to 100.00GB
12/01/2010 23:00:57 SETI@home Restarting task 27dc06ab.4655.2117.16.10.147_1 using setiathome_enhanced version 609


btw, i have been playing with the "additional work buffer" at Preferences... May it be the cause?

oh, and, the "restarted WU" shown in the last line, is restarted using the GPU, and none of the CPU cores are used :)

btw, i have been playing with the "additional work buffer" at Preferences... May it be the cause?

Yes, it could be related - when you make changes on that page, every single setting on all three 'Preferences' pages get read and acted on. Look and see whether you've perhaps unticked "Do work ... while computer is in use" (processor usage tab).

oh, and, the "restarted WU" shown in the last line, is restarted using the GPU, and none of the CPU cores are used :)

Yes, I can see from the 'version 609' at the end :-)

12/01/2010 23:00:56 Reading preferences override file

And check your local preferences (Advanced/Preferences...), as they override the online ones. If you only want to use the online preferences, click the Clear button at the local preferences window.

Gruß,
Gundolf

12/01/2010 23:00:56 SETI@home Computer location: home
12/01/2010 23:00:56 SETI@home General prefs: no separate prefs for home; using your defaults

And check your Setiathome Project preferences, that you haven't got the default preferences set to not get CPU work,
as you haven't set up Seti Project Preferences for the Home Venue.

Claggy

Edit: with Boinc 6.10.18 and earlier, if you turn off CPU or GPU work fetch in the Default project Preferences, it also Blocks work fetch for computers in the Home/School/Work venues as well,
(fixed in Boinc 6.10.19)

Look and see whether you've perhaps unticked "Do work ... while computer is in use" (processor usage tab).

Hi, no, i dont have any un-ticked option there....

i have checked and restarted BOINC several times, and i dont find a solution.. :-/

The thing is, cores start working when i re-fetch WUs from server, but after analyzing several of them (half of them, i would say), CPU stops....

Ah-ha,
i have switched to another project, docking@home, and cores work correctly

so its a seti@home specific preferences issue, lets investigate in that way :)

nah, i have reviewed many times all Preferences pages... they are identical for both projects
i have no idea about whats wrong with it...

Also i have removed the "home" preferences, so that now i only have the default ones.... but they are identical in both seti@home and docking@home

and docking@home uses the cpu... seti@home doesnt....

Look and see whether you've perhaps unticked "Do work ... while computer is in use" (processor usage tab).

Hi, no, i dont have any un-ticked option there....

i have checked and restarted BOINC several times, and i dont find a solution.. :-/

The thing is, cores start working when i re-fetch WUs from server, but after analyzing several of them (half of them, i would say), CPU stops....

After looking at your hosts task list, you're completed all your CPU tasks, leaving only GPU tasks left,
Is Boinc asking for CPU work from Seti?, since your Boinc log didn't include any work requests at all.

Claggy

Edit: after another look at your host, your GPU has been doing VLAR (very low angle range) tasks which are very slow using the Cuda app,
which will have driven your host's DCF (Duration Correction Facture) up, Boinc will have thought that all Seti tasks were going to take longer,
and so wouldn't have received as much CPU work, once you're finished any VLAR tasks on the GPU, and the GPU starts doing non-VLAR tasks,
then DCF will go down, run times wiil look more reasonable, and you'll get more CPU work.

hostid=5257895

After looking at your hosts task list, you're completed all your CPU tasks, leaving only GPU tasks left,

hm.. how can i know which are GPU tasks and which CPU tasks?

Is Boinc asking for CPU work from Seti?, since your Boinc log didn't include any work requests at all.

Well, i stopped BOINC to automatically fetch more WUs, since i have many of them available locally to analyse, let me show you a shot about my current BOINC tasks

http://img695.imageshack.us/i/boinc.jpg/

http://img695.imageshack.us/i/boinc.jpg/

I made the link active, though the image seems to be deleted already.

hm.. how can i know which are GPU tasks and which CPU tasks?

As Richard Haselgrove wrote in his Message 30508:

oh, and, the "restarted WU" shown in the last line, is restarted using the GPU

Yes, I can see from the 'version 609' at the end :-)

The CPU application (currently) has a '603' in the name.

Gruß,
Gundolf

---

Computer sind nicht alles im Leben. (Kleiner Scherz)

Edit: after another look at your host, your GPU has been doing VLAR (very low angle range) tasks which are very slow using the Cuda app,
which will have driven your host's DCF (Duration Correction Facture) up, Boinc will have thought that all Seti tasks were going to take longer,
and so wouldn't have received as much CPU work, once you're finished any VLAR tasks on the GPU, and the GPU starts doing non-VLAR tasks,
then DCF will go down, run times wiil look more reasonable, and you'll get more CPU work.

So, finally, you think this is the reason?

and btw, link is not working anymore.. sorry :(

Edit: after another look at your host, your GPU has been doing VLAR (very low angle range) tasks which are very slow using the Cuda app,
which will have driven your host's DCF (Duration Correction Facture) up, Boinc will have thought that all Seti tasks were going to take longer,
and so wouldn't have received as much CPU work, once you're finished any VLAR tasks on the GPU, and the GPU starts doing non-VLAR tasks,
then DCF will go down, run times wiil look more reasonable, and you'll get more CPU work.

So, finally, you think this is the reason?

and btw, link is not working anymore.. sorry :(

Yes, Basically because of the above, you ran out of CPU work, and Boinc couldn't ask for more work, eithier because Seti was set to NNT, Network Activity was Suspended,
or Boinc just didn't have a Internet connection.

Claggy

eithier because Seti was set to NNT

yeah, i set it to not download more WUs until i finish the ones i have in queue... but i didnt see that those WUs could onle be processed by the GPU

ok, thank you so much for your help :)