Message boards :
Number crunching :
BOINC V6.6.36 BUGs
Message board moderation
Author | Message |
---|---|
Sutaru Tsureku Send message Joined: 6 Apr 07 Posts: 7105 Credit: 147,663,825 RAC: 5 |
I had let run BOINC V6.6.36.. Two days well.. after some time I reached a ~ 4 day WU cache.. ~ 3,500 WUs.. And then.. surprise..? ;-) ..no.. :-( Normally every CUDA task [AR 0.44x WU] get ~ 95 MB system RAM support. Normally ~ 700 MB system RAM usage whole PC. [2 x 1 GB installed] Then BOINC V6.6.36 get panic and gone in EDF mode. [EarliestDeadlineFirst] Or something like this.. because.. ..but.. there were WUs with earlier deadline in BOINC. After and after he started and finished a lot of WUs. BOINC started and some sec. before the WU would normally finish - the WU suspended and a new WU start. But.. then.. I had ~ 20 suspended WUs. Every suspended CUDA task used RAM and finally the whole system RAM was full. Also the CPU was 100 % used. And this is an only GPU cruncher! No tasks on CPU! Normally every CUDA task get max. ~ 6 % CPU [~ 24 % Core] support. But now the active CUDA tasks get sometimes [every ~ 5 sec.] ~ 1 % CPU [~ 4 % Core] support. So no CUDA crunching possible. And this happened although: Leave applications in memory while suspended? NO [EDIT: BTW. A reboot didn't helped.. the suspended tasks used the system RAM again. I needed to abort this WUs.] So.. SETI@home community.. :-)
|
Sutaru Tsureku Send message Joined: 6 Apr 07 Posts: 7105 Credit: 147,663,825 RAC: 5 |
Ahh.. yes.. ..also I had two 'Falling back to HOST CPU'..: [...] Work Unit Info: ............... WU true angle range is : 0.202289 Cuda error 'cudaMalloc((void**) &dev_WorkData' in file 'd:/BTR/SETI6/SETI_MB_CUDA/client/cuda/cudaAcceleration.cu' in line 293 : out of memory. setiathome_CUDA: CUDA runtime ERROR in device memory allocation (Step 1 of 3). Falling back to HOST CPU processing... Flopcounter: 26974829277152.035000 [...] and [...] setiathome_CUDA: CUDA Device 1 specified, checking... Device 1: GeForce GTX 260 is okay SETI@home using CUDA accelerated device GeForce GTX 260 V10 modification by Raistmer Priority of worker thread rised successfully Priority of process adjusted successfully Total GPU memory 0 free GPU memory 0 Cuda error 'cufftPlan1d(&fft_analysis_plans[FftNum], FftLen, CUFFT_C2C, NumDataPoints / FftLen)' in file 'd:/BTR/SETI6/SETI_MB_CUDA/client/cuda/cudaAcc_fft.cu' in line 49 : no CUDA-capable device is available. Cuda error 'cufftPlan1d(&fft_analysis_plans[FftNum], FftLen, CUFFT_C2C, NumDataPoints / FftLen)' in file 'd:/BTR/SETI6/SETI_MB_CUDA/client/cuda/cudaAcc_fft.cu' in line 49 : no CUDA-capable device is available. setiathome_CUDA: CUDA runtime ERROR in plan FFT. Falling back to HOST CPU processing... setiathome_enhanced 6.02 Visual Studio/Microsoft C++ [...] ..but this happened ~ one day before I reached the ~ 4 day cache. |
1mp0£173 Send message Joined: 3 Apr 99 Posts: 8423 Credit: 356,897 RAC: 0 |
If you want to report problems with BOINC you really should report them to the BOINC developers and not to SETI@Home. SETI@Home is a science project that uses BOINC. |
Sutaru Tsureku Send message Joined: 6 Apr 07 Posts: 7105 Credit: 147,663,825 RAC: 5 |
You mean in the BOINC forum? http://boinc.berkeley.edu/dev AFAIK, the devs don't look there. And the user frequency is very very low there. Here would be the frequency of people bigger which have maybe also found this BUGs. Also.. hmm.. no thanks - I'll never again open a ticket.. My only one was deleted because of for me don't understood reason.. |
Raistmer Send message Joined: 16 Jun 01 Posts: 6325 Credit: 106,370,077 RAC: 121 |
The devs look there, be sure. BOINC problems better report to this board: http://boinc.berkeley.edu/dev/forum_forum.php?id=10 And situation you described really looks as new nasty bug in BOINC so it's really worth to report on BOINC forum. |
[AF>france>pas-de-calais]symaski62 Send message Joined: 12 Aug 05 Posts: 258 Credit: 100,548 RAC: 0 |
oué ^^ NVIDIA 185.85 version & CUDA 2.2 version SETI@Home Informational message -9 result_overflow with a general handicap of 80% and it makes much d' efforts for the community and s' expimer, thank you d' to be understanding. |
Jord Send message Joined: 9 Jun 99 Posts: 15184 Credit: 4,362,181 RAC: 3 |
The devs look there, be sure. The devs hardly look there. They usually only come over for a quick snoop when one of us emails them. You can better email the boinc_dev email list, which is a more direct line to them and then in a clear and concise way, with evidence, logs and all that, explain that you have found a bug. The list needs registration. |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14650 Credit: 200,643,578 RAC: 874 |
The devs look there, be sure. I find it works best if you restrict yourself to one bug per email, and help them to help you by putting all the relevant information in that one email. |
Hammeh Send message Joined: 21 May 01 Posts: 135 Credit: 1,143,316 RAC: 0 |
Yes - agreed with above. Bugs reported on the mailing list are normally investigated by the devs. To provide all of the needed information you need to run BOINC with <work_fetch_debug>, <sched_op_debug> and one run with <debt_debug> from the cc_config.xml file. Catch that log and post with it on the BOINC dev mailing list, putting the log into the email. Attachments to the list will be deleted by the email server. This may help the devs identify and fix the bug you have found. If you don't include the logs above, they will probably reply and tell you to run them anyway! |
-= Vyper =- Send message Joined: 5 Sep 99 Posts: 1652 Credit: 1,065,191,981 RAC: 2,537 |
The devs look there, be sure. Touche, at worst i had around 200 suspended ones whereas Sutaru had 20. I run this version atm but before this i needed 6.4.7 because it didn't EDF at all. Regards Vyper _________________________________________________________________________ Addicted to SETI crunching! Founder of GPU Users Group |
©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.