AK V8 + CUDA MB team work mod

[ofp1]

Link on your host, please ?

http://setiathome.berkeley.edu/show_host_detail.php?hostid=4711443

It seems is receiving WU's  now, I 'll check when I get home, but it still don't recognise the GTX260, yesterday it was OK
Thanks

[Raistmer]

This result came from CUDA MB app.
http://setiathome.berkeley.edu/result.php?resultid=1159848143

BTW, update both CPU and GPU parts of package (links buried iun thread ~2-3 pages back) there is new apps available.
Maybe I do repack soon (don't want to touch it before opt AP new release)

[perryjay]

Matt L posted this just a few minutes ago 

"Before releasing the astropulse application Eric had to add a couple fields to the result tables in the science database that are now necessary. These are large fields, and it's taking informix forever to update the table. The job was started 24 hours ago and is still chugging along. I guess it doesn't help that the assimilator queue is still rather large (though it is draining). So the release is delayed until this job finishes."

Looks like it won't be long now.

[ofp1]

This result came from CUDA MB app.
http://setiathome.berkeley.edu/result.php?resultid=1159848143

BTW, update both CPU and GPU parts of package (links buried iun thread ~2-3 pages back) there is new apps available.
Maybe I do repack soon (don't want to touch it before opt AP new release)

Hi,
I found the problem , when reinstalled Boinc 6.4.5 it has the option "Run in Protected Mode" and that is imcompatible with CUDA, just reinstall it again and uncheck the protected mode, now itīs crunching again with GPU

Iīll wait your repack of the new release in order to update all.

Thanks

[perryjay]

This from Eric Korpela:

FYI, Astropulse 5.03 was released and workunit generation is back on. However, because the results are not compatible with 5.02, it was added as a new application (astropulse_v5).

Is the new op-ap ready yet?? 

[Josef W. Segur]

Real Soon Now...

[Raistmer]

Is the new op-ap ready yet?? 

Just after it passes internal validation tests. Consider it as "ready"

[Richard Haselgrove]

I found the problem , when reinstalled Boinc 6.4.5 it has the option "Run in Protected Mode" and that is imcompatible with CUDA, just reinstall it again and uncheck the protected mode, now itīs crunching again with GPU

For the benefit of other readers: CUDA + Protected Mode is fine for Windows XP: it's just Vista (and I guess Windows 7) where you have to avoid Protected Mode.

[Archangel999]

Jason G you are absolutly right i use the app that Raimster said and now only 1 wokrunit is in idle i think this is because of cuda use 4% but can i do to work all the 5 


Best Regards
D.Draganov

[Jason G]

because of cuda use 4% but can i do to work all the 5 

I have a dumb idea ...  What happens if you set the cuda app info section to say it uses 0.000000 CPUs ?

[Edit: But probably wouldn't try it with the 'team' version .. maybe you'd get many many instances ? ]

[Raistmer]

team version doesn't sencible to CUDA plan at all.
BOINC doesn't know that it runs CUDA app in "team" combo
It matters only for V7

[Archangel999]

hm will see what have with the newer pack hope will fix it 

[WimTea]

This result s my first with v8b and it's a bit weird. Found CUDA device but reverted back to CPU after encountering an error:
Cuda error 'cufftPlan1d(&fft_analysis_plans[FftNum], FftLen, CUFFT_C2C, NumDataPoints / FftLen)' in file 'd:/BTR/seticuda/seti_boinc/client/cuda/cudaAcc_fft.cu' in line 49 : out of memory.
setiathome_CUDA: CUDA runtime ERROR in plan FFT. Falling back to HOST CPU processing...

Any ideas? QX9550 11% OC'd, XP 64, 4Gb RAM, PNY Geforce 9800 GT 512 Mb, CUDA driver 181.22, SDK and toolkit 2.1 as described in Seti thread.
Will try another WU when some are available  again and report on that one.

[edit]
Just got another one. stderr.txt has the same message unforunately. Showed up about 10 sec. after starting the WU.
[/edit]

[Raistmer]

You have memory leak
Total GPU memory 536543232    free GPU memory 14418944
too low free GPU memory.
Try to restart OS. If it will happen again try another driver version. I had memory leak on Win2003 x64 too before. Now not, but it work less stable than Vista x86.
It seems x64 nVidia drivers too buggy still ....

[WimTea]

Tx for the analysis. I'll try a reboot and different drivers. Don't want to revert to XP 32 though... Btw could this leak also explain why every now and then the screen (an old faithful) turns black for a few secs?

[edit]
Had to try another one before a reboot:
Total GPU memory 0    free GPU memory 0
 
[/edit]
[edit2]
Rebooted. Mem OK now, but still no go...

Cuda error 'cudaMemcpyToSymbol(cudaAcc_GaussFit_settings, (void*) &settings, sizeof(settings))' in file 'd:/BTR/seticuda/seti_boinc/client/cuda/cudaAcc_gaussfit.cu' in line 454 : invalid device symbol.

</stderr_txt>
]]>
 
Validate state Invalid
Claimed credit 0.0649666881657264
Granted credit 0
application version 6.08

[/edit2]
[edit3]
Seems the GPU bios version is also an issue, it advertises as an engineering version... Back to the shop with it I guess.
[/edit3]