Problem with running GPU tasks

[k6xt]

 While SNAFU still reigns supreme I overcame the GPU stoppage. I noticed in Windows system tray that GPU computing was "disabled at user request". Wrong. Actually I'd set BOINC to 'use GPU while computer is in use' but that factoid was ignored.

Bull in the china shop, I deleted BOINC directories and started all over again. Presto! GPU computing is working. My regrets to all the inconvenienced wingmen. I plan to stick with 6.12.34(64) for a while. This all started when I installed BOINC 7.0.25(x64) and Nvidia 301.42.

----------------------------------------------------

Situation normal in the SNAFU sense. I just upgraded to Nvidia 301.42 and Boinc 7.0.25(x64). Suddenly my GPU is at idle, no GPU work although I have downloaded GPU work in my queue from Collatz and Primegrid. Its not running.

GPU Caps Viewer says the GPU core is idling along at 50MHz, load near zero. Here's a copy of Boinc startup event log. Any experts see a clue here?

PS - thinking "corruption" I reinstalled Nvidia, no improvement.

Thanks Art
6/6/2012 6:19:09 PM |  | Starting BOINC client version 7.0.25 for windows_x86_64
6/6/2012 6:19:09 PM |  | log flags: file_xfer, sched_ops, task
6/6/2012 6:19:09 PM |  | Libraries: libcurl/7.21.6 OpenSSL/1.0.0d zlib/1.2.5
6/6/2012 6:19:09 PM |  | Data directory: D:\boinc
6/6/2012 6:19:09 PM |  | Running under account Art
6/6/2012 6:19:09 PM |  | Processor: 4 GenuineIntel Intel(R) Core(TM)2 Quad CPU    Q9650  @ 3.00GHz [Family 6 Model 23 Stepping 10]
6/6/2012 6:19:09 PM |  | Processor: 6.00 MB cache
6/6/2012 6:19:09 PM |  | Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss htt tm pni ssse3 cx16 sse4_1 syscall nx lm vmx smx tm2 pbe
6/6/2012 6:19:09 PM |  | OS: Microsoft Windows 7: Ultimate x64 Edition, Service Pack 1, (06.01.7601.00)
6/6/2012 6:19:09 PM |  | Memory: 6.00 GB physical, 12.00 GB virtual
6/6/2012 6:19:09 PM |  | Disk: 238.24 GB total, 194.69 GB free
6/6/2012 6:19:09 PM |  | Local time is UTC -6 hours
6/6/2012 6:19:09 PM |  | NVIDIA GPU 0: GeForce GTX 560 Ti (driver version 301.42, CUDA version 4.20, compute capability 2.1, 1024MB, 914MB available, 1352 GFLOPS peak)
6/6/2012 6:19:09 PM |  | OpenCL: NVIDIA GPU 0: GeForce GTX 560 Ti (driver version 301.42, device version OpenCL 1.1 CUDA, 1024MB, 914MB available)
6/6/2012 6:19:09 PM | Milkyway@Home | Found app_info.xml; using anonymous platform
6/6/2012 6:19:09 PM | SETI@home | Found app_info.xml; using anonymous platform
6/6/2012 6:19:09 PM |  | Config: use all coprocessors
6/6/2012 6:19:09 PM | Collatz Conjecture | URL http://boinc.thesonntags.com/collatz/; Computer ID 19418; resource share 100
6/6/2012 6:19:09 PM | Einstein@Home | URL http://einstein.phys.uwm.edu/; Computer ID 5411865; resource share 1000
6/6/2012 6:19:09 PM | Milkyway@Home | URL http://milkyway.cs.rpi.edu/milkyway/; Computer ID 100603; resource share 100
6/6/2012 6:19:09 PM | SETI@home | URL http://setiathome.berkeley.edu/; Computer ID 5451708; resource share 100
6/6/2012 6:19:09 PM | PrimeGrid | URL http://www.primegrid.com/; Computer ID 170826; resource share 100
6/6/2012 6:19:09 PM |  | General prefs: from http://bam.boincstats.com/ (last modified 18-Jul-2011 09:02:19)
6/6/2012 6:19:09 PM |  | Computer location: home
6/6/2012 6:19:09 PM |  | General prefs: no separate prefs for home; using your defaults
6/6/2012 6:19:09 PM |  | Preferences:
6/6/2012 6:19:09 PM |  | max memory usage when active: 4607.37MB
6/6/2012 6:19:09 PM |  | max memory usage when idle: 5836.00MB
6/6/2012 6:19:09 PM |  | max disk usage: 100.00GB
6/6/2012 6:19:09 PM |  | (to change preferences, visit the web site of an attached project, or select Preferences in the Manager)
6/6/2012 6:19:09 PM |  | Not using a proxy

[Claggy]

Try installing Boinc 7.0.28 instead, 7.0.25 had a number of Bugs since fixed,

Claggy

[William]

Startup messages don't tell you much about running tasks. You would need to set a number of debug flags in cc_config.xml.
you want cpu_sched_debug and coproc_debug. you also want to attach the files global_prefs.xml and global_prefs_override.xml here.

What could have happened is that you looked at the local preferences and instead of 'cancel' clikced 'ok' thereby enableing [default] local preferences. These override the ones supplied via the web.
Depending on boinc version it can grab some of the settings from the actual web setting, making no difference at start, but making it impossible to change anything later via the web settings.
Deleting the boinc dirs would have deleted the local preferences files, thereby intruction boinc to use the web settings again.

NB if you edit a post at a later time because you want to add something it REALLY helps if you mark the new portions with 'edit'.

I take it the upper part of your post is the later edit, meaning it is working now.

You should still skip boinc 7.0.25 and update straight to boinc 7.0.28 if you want to update.

[k6xt]

Try installing Boinc 7.0.28 instead, 7.0.25 had a number of Bugs since fixed,

Claggy

Thanks for the tip. Reading between the lines, 28 is more stable than 25.  Although Boinc says its "for testing only". After deleting the directories and starting over its all working now with V6.2.34(x64).

Am I missing anything vital by sticking with it?
Thanks
Art

[Mike]

Not if you are using opti apps.
I´m still at 6.12.22. (modified)

Mike
 

[k6xt]

Not if you are using opti apps.
I´m still at 6.12.22. (modified)
Mike

Which I am, the latest KWSN installer. Thanks!

[Fredericx51]

Yep, it's a great installer v40 ,  although I still have problems with  MB WU on  ATI GPUs,
HD5870s.  Two of them, probably also part of the problem and the INTEL DP67BG mobo,
2 x PCIe (2.0) x 16 and x 8, when both used in 2 x 8 mode.  PSU = 1000Watt.

During the first 20 seconds 1 CPU- core uses 100% then GPU makes first step in %,  after this ~ 20-30% .

One  CPU core, 2 threads are used to load and unload the GPUs .
As a last resort, upped the base 100MHz clock to 102MHz to get some more FLOPS from
the CPU, multiplier is locked at 34 max. also rased GPUcore frequency from 850 to 876 MHz..
Working quit stable, hope it stops the Max Time Exeeded, errrors. 900MHz. will probably do OK.

Just noticed at 57:47 minutes the running WU (610) is stopped with the message Time exceeded, or something like it.
This only happens to a VLAR ! Mostly on the 2nd GPU, device 1.

The AstroPulse OpenCL app. r.1316 works very fast, a none blanked WU is done in less then
an hour, a >90% blanked WU takes 2 till 3 hour GPU and  1 to 2 hours on  CPU .
(CPU = I7-2600; GPUs = 2x HD5870) Haven't tried rev1363 yet.

And the SSE3/AVX capable rev.331 (339?) for  CPU is very efficient,  too.
(Wonder if a newer version for MB work is to be expected)? 

[Raistmer]

Wonder if a newer version for MB work is to be expected? 

Yes, it's expected.

[Fredericx51]

Wonder if a newer version for MB work is to be expected? 

Yes, it's expected.

Great , can't wait