brp4cuda32nv301 fails instantly
5 Jul 2012 13:33:36 UTC
Topic 196440
(moderation:
I have just installed BOINC on a new i73770K with windows 7 pro 64 bit, 32Gb RAM, BOINC 7.0.28 (64bit)
For some reason the BRP work units for NVIDIA CUD
A fail on start.
Driver Version: 301.42 and upgraded to beta 304.79
Medusa
41 Einstein@Home 5/07/2012 10:48:50 PM Starting task b2030.20110505.G39.65+00.81.N.b3s0g0.00000_3728_0 using einsteinbinary_BRP4 version 125 (BRP4cuda32nv301) in slot 12
42 Einstein@Home 5/07/2012 10:48:54 PM Computation for task b2030.20110505.G39.65+00.81.N.b3s0g0.00000_3728_0 finished
43 Einstein@Home 5/07/2012 10:48:54 PM Output file b2030.20110505.G39.65+00.81.N.b3s0g0.00000_3728_0_0 for task b2030.20110505.G39.65+00.81.N.b3s0g0.00000_3728_0 absent
44 Einstein@Home 5/07/2012 10:48:54 PM Output file b2030.20110505.G39.65+00.81.N.b3s0g0.00000_3728_0_1 for task b2030.20110505.G39.65+00.81.N.b3s0g0.00000_3728_0 absent
45 Einstein@Home 5/07/2012 10:48:54 PM Output file b2030.20110505.G39.65+00.81.N.b3s0g0.00000_3728_0_2 for task b2030.20110505.G39.65+00.81.N.b3s0g0.00000_3728_0 absent
46 Einstein@Home 5/07/2012 10:48:54 PM Output file b2030.20110505.G39.65+00.81.N.b3s0g0.00000_3728_0_3 for task b2030.20110505.G39.65+00.81.N.b3s0g0.00000_3728_0 absent
47 Einstein@Home 5/07/2012 10:48:54 PM Output file b2030.20110505.G39.65+00.81.N.b3s0g0.00000_3728_0_4 for task b2030.20110505.G39.65+00.81.N.b3s0g0.00000_3728_0 absent
48 Einstein@Home 5/07/2012 10:48:54 PM Output file b2030.20110505.G39.65+00.81.N.b3s0g0.00000_3728_0_5 for task b2030.20110505.G39.65+00.81.N.b3s0g0.00000_3728_0 absent
49 Einstein@Home 5/07/2012 10:48:54 PM Output file b2030.20110505.G39.65+00.81.N.b3s0g0.00000_3728_0_6 for task b2030.20110505.G39.65+00.81.N.b3s0g0.00000_3728_0 absent
50 Einstein@Home 5/07/2012 10:48:54 PM Output file b2030.20110505.G39.65+00.81.N.b3s0g0.00000_3728_0_7 for task b2030.20110505.G39.65+00.81.N.b3s0g0.00000_3728_0 absent
7.0.28
- exit code -1073741515 (0xc0000135)
]]>
Other CUDA BOINC projects work fine and all Einstein CPU apps also work fine.
I tried to disable the BRP application on this new computer by assigning it to location=work and restricting the applications but my BOINC client doesn't recognise my separate prefs for "work".
21 Einstein@Home 5/07/2012 10:48:01 PM General prefs: from Einstein@Home (last modified 22-Jan-2011 12:26:27)
22 Einstein@Home 5/07/2012 10:48:01 PM Computer location: work
23 Einstein@Home 5/07/2012 10:48:01 PM General prefs: no separate prefs for work; using your defaults
Any Ideas?
brp4cuda32nv301 fails instantly
)
When you edited your preferences on the web site, there is a clear message on top in red saying
Your cited client messages ("last modified 22-Jan-2011") show that this hasn't happened (yet).
BM
BM
As for the CUDA tasks
)
As for the CUDA tasks crashing there is another forum thread exit code -1073741515 (0xc0000135) discussing this problem.
BM
BM
RE: RE: Einstein@Home 5/0
)
The message showing is for updates on the computing preferences, not for updating the project preferences.
I've changed the Einstein and Seti project preferences a lot in the past weeks, but my client message on the prefs still says "SETI@home | General prefs: from SETI@home (last modified 29-Feb-2012 07:17:27)", which as far as I know is correct. I did change something in the online global preferences in February, but use the local computing preferences for the rest of the time.
Hm. Do you need to set up
)
Hm. Do you need to set up global preferences for a venue in order to make project preferences for that venue work?
BM
BM
RE: Hm. Do you need to set
)
No.
(although the only project I can confirm that for explicity is SETI. But I do have a test account there which has only the default global preference, but both default and 'work' project preferences. 'Work' works.)
RE: RE: Hm. Do you need
)
I have no venues defined for global preferences (It was hard to find a common setting for them all so I do use the local settings to set each host global settings).
But my hosts are assigned to different venues on Einstein and SETI for the project prefferences and it works for both.
Indeed, Im using different project venues, so one host that is "home" (backup 0 resource share) for SETI is also "work" (normal resouce share) for Eintein and viceversa.
RE: RE: Einstein@Home 5/0
)
The message showing is for updates on the computing preferences, not for updating the project preferences.
Good to know that the message refers to the computing preferences
Even so the client is still downloading BRP workunits even after several forced updates.
NOTE Einstein preferences for "work":
Run only the selected applications Binary Radio Pulsar Search (Arecibo): no
Gravitational Wave S6 LineVeto search: yes
Gamma-ray pulsar search #1: yes
If no work for selected applications is available, accept work from other applications? no
I will try adding a separate computing preference for work and see if the client picks up the preference for the project.
RE: RE: RE: Einstein@Ho
)
I think Ive read that there was some issue setting the apps prefferences in that under certain circumstances the settings will not be enforced...
IIRC, you may need to enable/disable all the apps first (forcing the update in the client in the process) and then disable the unwanted apps...
You can test if Boinc is reading the prefferences just changing the resource share number... If the new share number is noticed on your "work" hosts, and still they are downloading BRP tasks, you can try to disable the "Use NVidia GPU" setting (and disable the Run CPU version for wich GPU apps are available if you dont want to crunch BRPs neither on the CPU)