Computation error Intel GPU

Questions and Answers : Windows : Computation error Intel GPU
Message board moderation

To post messages, you must log in.

AuthorMessage
djoser

Send message
Joined: 4 May 00
Posts: 9
Credit: 628,738
RAC: 0
Germany
Message 1722772 - Posted: 6 Sep 2015, 9:12:57 UTC

Hello everyone!

Yesterday i set up a Windows 8.1 64-bit machine dedicated for boinc.
On my first two gpu-jobs from seti i got this results:

http://setiathome.berkeley.edu/result.php?resultid=4364078300
http://setiathome.berkeley.edu/result.php?resultid=4364078138

Additional information: I'm running one vlhc@home task on one cpu simultanously.
Seti is configured to use the gpu only.

What went wrong and how can i fix it, because the current job seems to have the same problem.

I would gladly post this in the number crunching forum, but i don't have enough credit to start a thread there!?!

Thanks in advance,
djoser.
ID: 1722772 · Report as offensive
Profile BilBg
Volunteer tester
Avatar

Send message
Joined: 27 May 07
Posts: 3720
Credit: 9,385,827
RAC: 0
Bulgaria
Message 1723004 - Posted: 7 Sep 2015, 2:49:03 UTC - in response to Message 1722772.  

Exit status 197 (0xc5) EXIT_TIME_LIMIT_EXCEEDED
exceeded elapsed time limit 19474.11 (1414546.15G/72.64G)

Obviously BOINC estimates are too low.
BOINC thinks the tasks will finish in 16 minutes and they were running for 324 minutes


The next is a fix for Advanced user - do it on your own risk:

- Exit BOINC
- in BOINC Data directory find the file client_state.xml
- make Copy of it (to start over if you made mistake)

- Edit client_state.xml by Notepad and:

Ctrl+H

Find what:
<rsc_fpops_bound>

Replace with:
<rsc_fpops_bound>1

[Replace All]

Ctrl+S

- Start BOINC


The above will make BOINC wait ~3000 minutes before aborting task
After some (>10) tasks are completed BOINC will learn the speed of your GPU and next Downloads will have more good estimates.

If you don't understand what I wrote - don't do it!
 


- ALF - "Find out what you don't do well ..... then don't do it!" :)
 
ID: 1723004 · Report as offensive
Profile Jord
Volunteer tester
Avatar

Send message
Joined: 9 Jun 99
Posts: 15184
Credit: 4,362,181
RAC: 3
Netherlands
Message 1723161 - Posted: 7 Sep 2015, 15:30:08 UTC - in response to Message 1723004.  

Exit status 197 (0xc5) EXIT_TIME_LIMIT_EXCEEDED
exceeded elapsed time limit 19474.11 (1414546.15G/72.64G)

Obviously BOINC estimates are too low.

Seeing as how this person's GPU is now crunching work normally, I feel it was a problem with initialization of the executable, also because of the error that showed further down the stderr.txt in his first task:

OpenCL-kernels filename : MultiBeam_Kernels_r2929.cl 
INFO: can't open binary kernel file: C:\ProgramData\BOINC/projects/setiathome.berkeley.edu\MultiBeam_Kernels_r2929.cl_IntelRHDGraphics.bin_V7_1018104226, continue with recompile...
Info : Building Program (binary, clBuildProgram):main kernels: OK code 0
INFO: binary kernel file created
WARNING: can't open binary kernel file for oclFFT plan: C:\ProgramData\BOINC/projects/setiathome.berkeley.edu\MB_clFFTplan_IntelRHDGraphics_524288_gr64_lr8_wg64_tw3_ls1024_bn16_cw16_r2929.bin_1018104226, continue with recompile...
WARNING: can't open binary kernel file for oclFFT plan: C:\ProgramData\BOINC/projects/setiathome.berkeley.edu\MB_clFFTplan_IntelRHDGraphics_8_gr64_lr8_wg64_tw3_ls1024_bn16_cw16_r2929.bin_1018104226, continue with recompile...
WARNING: can't open binary kernel file for oclFFT plan: C:\ProgramData\BOINC/projects/setiathome.berkeley.edu\MB_clFFTplan_IntelRHDGraphics_16_gr64_lr8_wg64_tw3_ls1024_bn16_cw16_r2929.bin_1018104226, continue with recompile...
WARNING: can't open binary kernel file for oclFFT plan: C:\ProgramData\BOINC/projects/setiathome.berkeley.edu\MB_clFFTplan_IntelRHDGraphics_32_gr64_lr8_wg64_tw3_ls1024_bn16_cw16_r2929.bin_1018104226, continue with recompile...
WARNING: can't open binary kernel file for oclFFT plan: C:\ProgramData\BOINC/projects/setiathome.berkeley.edu\MB_clFFTplan_IntelRHDGraphics_64_gr64_lr8_wg64_tw3_ls1024_bn16_cw16_r2929.bin_1018104226, continue with recompile...
WARNING: can't open binary kernel file for oclFFT plan: C:\ProgramData\BOINC/projects/setiathome.berkeley.edu\MB_clFFTplan_IntelRHDGraphics_128_gr64_lr8_wg64_tw3_ls1024_bn16_cw16_r2929.bin_1018104226, continue with recompile...
WARNING: can't open binary kernel file for oclFFT plan: C:\ProgramData\BOINC/projects/setiathome.berkeley.edu\MB_clFFTplan_IntelRHDGraphics_256_gr64_lr8_wg64_tw3_ls1024_bn16_cw16_r2929.bin_1018104226, continue with recompile...
WARNING: can't open binary kernel file for oclFFT plan: C:\ProgramData\BOINC/projects/setiathome.berkeley.edu\MB_clFFTplan_IntelRHDGraphics_512_gr64_lr8_wg64_tw3_ls1024_bn16_cw16_r2929.bin_1018104226, continue with recompile...
WARNING: can't open binary kernel file for oclFFT plan: C:\ProgramData\BOINC/projects/setiathome.berkeley.edu\MB_clFFTplan_IntelRHDGraphics_1024_gr64_lr8_wg64_tw3_ls1024_bn16_cw16_r2929.bin_1018104226, continue with recompile...
WARNING: can't open binary kernel file for oclFFT plan: C:\ProgramData\BOINC/projects/setiathome.berkeley.edu\MB_clFFTplan_IntelRHDGraphics_2048_gr64_lr8_wg64_tw3_ls1024_bn16_cw16_r2929.bin_1018104226, continue with recompile...
WARNING: can't open binary kernel file for oclFFT plan: C:\ProgramData\BOINC/projects/setiathome.berkeley.edu\MB_clFFTplan_IntelRHDGraphics_4096_gr64_lr8_wg64_tw3_ls1024_bn16_cw16_r2929.bin_1018104226, continue with recompile...
WARNING: can't open binary kernel file for oclFFT plan: C:\ProgramData\BOINC/projects/setiathome.berkeley.edu\MB_clFFTplan_IntelRHDGraphics_8192_gr64_lr8_wg64_tw3_ls1024_bn16_cw16_r2929.bin_1018104226, continue with recompile...
WARNING: can't open binary kernel file for oclFFT plan: C:\ProgramData\BOINC/projects/setiathome.berkeley.edu\MB_clFFTplan_IntelRHDGraphics_16384_gr64_lr8_wg64_tw3_ls1024_bn16_cw16_r2929.bin_1018104226, continue with recompile...
WARNING: can't open binary kernel file for oclFFT plan: C:\ProgramData\BOINC/projects/setiathome.berkeley.edu\MB_clFFTplan_IntelRHDGraphics_32768_gr64_lr8_wg64_tw3_ls1024_bn16_cw16_r2929.bin_1018104226, continue with recompile...
WARNING: can't open binary kernel file for oclFFT plan: C:\ProgramData\BOINC/projects/setiathome.berkeley.edu\MB_clFFTplan_IntelRHDGraphics_65536_gr64_lr8_wg64_tw3_ls1024_bn16_cw16_r2929.bin_1018104226, continue with recompile...
WARNING: can't open binary kernel file for oclFFT plan: C:\ProgramData\BOINC/projects/setiathome.berkeley.edu\MB_clFFTplan_IntelRHDGraphics_131072_gr64_lr8_wg64_tw3_ls1024_bn16_cw16_r2929.bin_1018104226, continue with recompile...
ar=2.726646  NumCfft=99655  NumGauss=0  NumPulse=27847832376  NumTriplet=27847832376
Currently allocated 120 MB for GPU buffers

His second task didn't have that error, but showed abnormal values for autocorrection and spikes.

All the rest of his tasks run normal. Three of those finished tasks have the same AR as that of the faulty tasks, so should have the same flops value. Showing that isn't the problem.
ID: 1723161 · Report as offensive
djoser

Send message
Joined: 4 May 00
Posts: 9
Credit: 628,738
RAC: 0
Germany
Message 1723275 - Posted: 7 Sep 2015, 19:07:52 UTC - in response to Message 1723161.  

Hi everyone!

Thanks for taking the time and looking into my problem.

After my first two jobs failed after a fresh installation of Win and Boinc i paniced a little bit. But guess what: a simple reboot of the machine solved the problem. Since then the WUs are trouble free. Very slow of course, but what can you expect from the smallest iGPU Intel ever made...

Although i have to mention that after setup of boinc the machine was rebooted more than once before it began with crunching. The vlhc workunits were processed correctly right from the start, but they work in a separate virtual machine.

Still don't know why there was something wrong with boinc trying to open files.

Anyways, thanks again for your help!
Regards, djoser.
ID: 1723275 · Report as offensive
Profile BilBg
Volunteer tester
Avatar

Send message
Joined: 27 May 07
Posts: 3720
Credit: 9,385,827
RAC: 0
Bulgaria
Message 1723454 - Posted: 8 Sep 2015, 5:14:36 UTC - in response to Message 1723275.  

Still don't know why there was something wrong with boinc trying to open files.

If you mean this:
WARNING: can't open binary kernel file

1) This is Message from the app, Not from BOINC "trying to open files"

2) That is normal, those files don't exist at first run.
They are compiled (by the app + OpenCL driver) on first run from MultiBeam_Kernels_r????.cl
They are reused on next runs (to save compilation time)

They will be re-compiled when/if the app version or driver version changes.
 


- ALF - "Find out what you don't do well ..... then don't do it!" :)
 
ID: 1723454 · Report as offensive

Questions and Answers : Windows : Computation error Intel GPU


 
©2024 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.