Einstein - Berechnungsfehler
22 Aug 2013 18:08:11 UTC
Topic 197157
(moderation:
Hardware:
- NVIDIA GeForce GT640M-LE
- Intel HD Graphics 4000
- Intel i7 3632QM
- Win7 64Bit
Seit einiger Zeit kommen Jobs mit NVIDIA-Nutzung. Doch der Job wird sofort mit Berechnungsfehler abgebrochen.
Was mache ich falsch ?
Im Log wird kein Fehlerabbruch angezeigt, nur fertig gestellt in 2 Sekunden.
22.08.2013 16:37:59 | Einstein@Home | Starting task p2030.20130625.G55.19-00.45.S.b0s0g0.00000_1184_1 using einsteinbinary_BRP4G version 139 (BRP4G-cuda32-nv301) in slot 8
22.08.2013 16:38:01 | Einstein@Home | Computation for task p2030.20130625.G55.19-00.45.S.b0s0g0.00000_1184_1 finished
Einstein - Berechnungsfehler
)
You get error 0xc0000135 and that seems to be related to problems with the nvidia driver installation.
You are currently running version 306.97, the latest version is 320.49. I suggest trying to upgrade your nvidia driver.
Maybe removing the old driver completely before installing the new will help as well.
Vielen Dank für die schnelle
)
Vielen Dank für die schnelle Antwort. Ich installiere den neuen Treiber.
Hello I have got the same
)
Hello I have got the same Problem.
Environment is:
Gigabyte B75M-D3V with 8 Gig RAM
intel i7 3770 (@ stock frequency 3,4 GHz, not OC)
Gigabyte GeForce GTX 670 2 Gig @ stock frequency
OS Win 7 H.Prem. 64bit, SP1
Graphics driver 320.49 WHQL (latest)
BOINC-Manager 7.0.64 (x64)
Host Details: http://einsteinathome.org/host/6851171
Error: Binary Radio Pulsar Search says 'Berechnungsfehler' (computation error) after app. 0...2 seconds.
I am not quite shure if this problem appears in other E@H applications like Gamma-Ray or Gravitational Wave too (I cancelled computation after first 20 error msgs), but I will try to figure out.
Are there any known workarounds or 'safe' software combinations for this Problem ?
@admin The error code says
)
@admin
The error code says that your missing a DLL file.
Action:
Remove the driver.
Install the latest driver.
RE: @admin The error code
)
Yes, the problem is a missing DLL,
No, the solution isn't to reinstall the drivers,
the project supplies dll's for the BRP Cuda app, the solution is to manually download those dll's and overwrite the old versions in the project directory.
The files to try first are:
http://einstein2.aei.uni-hannover.de/download/cudart_xp32_32_16.dll
http://einstein2.aei.uni-hannover.de/download/cufft_xp32_32_16.dll
Claggy
Hi all, thank you for the
)
Hi all, thank you for the fast response.
I have set a system recovery point, uninstalled the driver and reinstalled it after a clean boot. Any positive effect is not noticeable.
Installed (and re-installed) driver was 320.49 WHQL (latest version as I mentioned before).
My questions regarding the DLLs: are they gonna work under Windows 64bit and: is there any official source of them at UWM or so? And - most important - where to put them ??
(aunty Edith wiped away the spelling errors - or most of them)
RE: My questions regarding
)
In the project directory as i already said, but since you haven't posted your Boinc startup messages i can't be sure what it is, but probably the same as Mcnef's in the adjoining thread.
Claggy
RE: In the project
)
sorry for overlooking your hint...
BOINC client is installed @ D:\Meine_Programme\BOINC
Project directory is D:\Meine_Programmdaten\BOINC\projects
where E@H client directory is located at subdir D:\Meine_Programmdaten\BOINC\projects\einstein.phys.uwm.edu
both DLLs are present @ this directory.
cudart_xp32_32_16.dll / date 2013-06-19 / ver. 6.14.11.3020 / 376 kB / CUDA runtime ver. 3.2.16
cufft_xp32_32_16.dll / date 2013-03-19 / ver. 6.14.11.3020 / 27,2 MB / CUDA FFT Library ver. 3.2.16
The files you recomended:
The files to try first are:
http://einstein2.aei.uni-hannover.de/download/cudart_xp32_32_16.dll
http://einstein2.aei.uni-hannover.de/download/cufft_xp32_32_16.dll
have the same version number and file size as the ones present.
It seems to me that there is a problem with E@H and BOINC 7.0.64(x64) at some SW constallations - I checked back with my results and I have a result for 'Gravitational Wave S6 Directed Search' which is accepted to be o.k. and is validated (Aug-24) under same HW & SW conditions and settings.
http://einsteinathome.org/host/6851171/tasks&offset=0&show_names=1&state=3&appid=24
Since the beginning of the
)
Since the beginning of the directed searches (CasA) most of the jobs errored out after 15 to 30 seconds. Only a few ones ran until the end. Some, which were running longer, errored out when the computer was rebooted.
Now the Perseus Arm Cuda jobs error out too, typically within 3 seconds.
The Gamma-ray pulsar search #2 v1.12 units are running ok, the Binary Radio Pulsar Search (Arecibo) v1.33 (BRP4X64) too, if I can get them.
I checked my System for viruses, for overheating problems, there ist noting to find.
The CPU is Core i7-3770K, overclocked to 3.7 GHz, OS Win 7 64 professional, the video card is NVIDIA GTX 650, driver version 320.49, Boinc Manager 7.0.64 (x 64), Antivirus is Kaspersky Internet Security 2014, all Boinc and Einstein applications set to "trusted".
I don't know what i should do to solve the problem(s), help is appreciated.
I had already tried Claggys
)
I had already tried Claggys hint - renewing the DLLs - with no success. Then i had some bluescreens with different messages and supposed a problem with my SSD or harddisk. After some testing i found out it was one RAM - module. Meanwhile the computer is running with only one RAM - module of 8 GB and i am trying to sort out the software damages caused by the RAM errors. The CasA workunits are running again, the Cudas continue to error out within some seconds.