I see since today that my backupproject on a poem machine is erroring out every einsteinunit.
http://einsteinathome.org/task/396515504
What does this error mean?
DSKAG Austria Research Team: [LINK]http://www.research.dskag.at[/LINK]
Copyright © 2024 Einstein@Home. All rights reserved.
Computing errors
)
Oh interesting after reset of the unattended computer it seem to work again O.o But there was no visible fault. I identified it after seeing that seti is set as second backupproject (didnt know that anymore) and failing too. So forget the thread ^^ Strange..
DSKAG Austria Research Team: [LINK]http://www.research.dskag.at[/LINK]
RE: I see since today that
)
I've the same issue since end of August...
http://einsteinathome.org/host/7703197/tasks&offset=20&show_names=1&state=0&appid=0
Seems my machine kills every E@H WU all with:
Stderr output
7.0.64
(unknown error) - exit code -1073741511 (0xc0000139)
]]>
Any clues on how to proceed from my end to fix this?
cheers!
Did you already reboot? Shut
)
Did you already reboot? Shut down, remove the power cord for 10+ minutes, then start again? What else have you already tried? What else is the machine running?
MrS
Scanning for our furry friends since Jan 2002
RE: Seems my machine kills
)
Your host isn't killing every Wu, only the Cuda GPU Wu's, your CPU Wu's seem to be unaffected,
Boinc now reports your GPU as having 188.22 drivers, this is well below Cuda32 capable, while previously the BRP Cuda app was reporting a Cuda55 driver, I suggest you try and your Nvidia Graphic drivers again, it'll need to be at least 266.xx
Claggy
RE: Boinc now reports your
)
ah-ja, and here is where the cat bites its tail...
The whole issue started once I'd joined the E@H project, with S@H alone all was running fine.
With my prev nVidia driver version frequently a small error message popped up , something like [nVidia driver stopped working]...
So I updated to their latest driver release, on which any GPU WU's instantly caused a BSOD... great...
As I can see only S@H WUs using the GPU, I inquired at the S@H forum...
OK, de-install all GPU drivers, de-install all DirectX utilities, build everything up new...
First I used the GPU driver offered by Windows auto install, which again cause the small [GPU driver quit working] warning...
So I used the OEM driver offered by the MFG of this machine; no fault messages there, no BSOD either, but apparently also no WUs functioning...
So if I install the latest version of nVidia GPU driver, I'll most likely have the BSOD again... cool...
RE: RE: Boinc now reports
)
You could try under-clocking/volting of the graphics card. Maybe it is a over-heating
RE: I've the same issue
)
Do you mean E@H was working previously?
MrS
Scanning for our furry friends since Jan 2002
Already did a hardware check
)
Already did a hardware check with OCCT, basically a waist of time, neither CPUs nor GPU exceed peaks of ~80°C, no faults, no issues...
GPU is also not overclocked, so I think there no issues with the hardware.
The machine is not left ON unattended, only in operation while I'm at home/working on it.
E@H did work at first after I'd joined (I'd given up on NEO to get online...), but then I did get the small system warnings about [GPU driver failed...], thus I went to the latest nVidia driver v320.78 which caused the BSOD some after the BOINC screensaver kicked in...
Currently I'm back to a prior version v188.22 and observing...
Thanx anyway, I'll report any further glitches.
cheers!
Your host scheduler log is
)
Your host scheduler log is telling you...
2013-09-03 19:15:54.6029 [PID=31787] [version] parsed project prefs setting 'gpu_util_brp': 0.000000
2013-09-03 19:15:54.6029 [PID=31787] [version] driver version required min: -26000, supplied: 18822
2013-09-03 19:15:54.6029 [PID=31787] [version] Checking plan class 'BRP5-cuda32-nv301'
2013-09-03 19:15:54.6029 [PID=31787] [version] parsed project prefs setting 'gpu_util_brp': 0.000000
2013-09-03 19:15:54.6029 [PID=31787] [version] driver version required min: -30100, supplied: 18822
that the version of driver you have supplied will not run the BRP4 or BRP5 GPU apps. You have to find and install a WHQL driver version greater than 301.00 to run both these apps. It has been this way for at least a year...
Gord
320.78 is an "enterprise"
)
320.78 is an "enterprise" driver - which may mean they're doing strange things there. Can you install a regular driver?
MrS
Scanning for our furry friends since Jan 2002