No Suitable CUDA device available

Forums › Cruncher's Corner

David Anderson
David Anderson
Joined: 3 Nov 10
Posts: 14
Credit: 123314944
RAC: 0
11 Aug 2016 16:27:08 UTC
Topic 199885
(moderation:
    )

    Once a day or so I get

    [01:54:55][21177][INFO ] Application startup - thank you for supporting Einstein@Home!
    [01:54:55][21177][INFO ] Starting data processing...
    GPU device # not found in init_data.xml or command line
    [01:54:55][21177][ERROR] No suitable CUDA device available!
    [01:54:55][21177][ERROR] Demodulation failed (error: 1001)!
    01:54:55 (21177): called boinc_finish(1001)

    An example from
    Task 570735312
    Name: PM0147_02961_150_1
    Work unit ID: 252012557

    Ubuntu 16.04 32bit , upgraded from 14.04 a few days ago.
    This is BRP6.
    Perhaps all BRP6 now failing on this particular old CPU/GPU?

    BRP4G CUDA do run ok on the machine.
    Computer: 3638092

    Suggestions? Is this my fault somehow?

    AgentB
    AgentB
    Joined: 17 Mar 12
    Posts: 915
    Credit: 513211304
    RAC: 0

    Two suggestions... Did you

    11 Aug 2016 20:41:28 UTC
    Message 148825
    (moderation:
      )

      Two suggestions...

      Did you have a good reason to leave 14.04?  For older hardware 14.04  might be a better place to stay.

      Nvidia driver: 304.13 - is that the latest working version?

      Good luck.

      David Anderson
      David Anderson
      Joined: 3 Nov 10
      Posts: 14
      Credit: 123314944
      RAC: 0

      Sorry, not a good reason to

      12 Aug 2016 0:22:34 UTC
      Message 148834
      (moderation:
        )

        Sorry, not a good reason to leave 14.04.  Just keeping the machines on the same release and was updating the others, so ...

        304.131 was what 16.04 chose. I just now went to the latest

        available, 340.96.   I hope the new driver helps.

        WB8ILI
        WB8ILI
        Joined: 20 Feb 05
        Posts: 45
        Credit: 1142969136
        RAC: 1560356

        David Anderson -  Did 340.96

        12 Aug 2016 19:01:36 UTC
        Message 148851
        (moderation:
          )

          David Anderson - 

          Did 340.96 solve your problem?

          Do you have WINE installed on that machine?

           

           

           

          mmonnin
          mmonnin
          Joined: 29 May 16
          Posts: 292
          Credit: 3444726540
          RAC: 799183

          Looking at their info it

          13 Aug 2016 0:42:47 UTC
          Message 148856
          (moderation:
            )

            Looking at their info it looks like it didn't. A 9500GT is spitting out a lot of tasks with computation errors.

            https://einsteinathome.org/task/570989650

            <core_client_version>7.6.31</core_client_version>
<![CDATA[
<message>
process exited with code 252 (0xfc, -4)
</message>
<stderr_txt>
[17:25:55][29555][INFO ] Application startup - thank you for supporting Einstein@Home!
[17:25:55][29555][INFO ] Starting data processing...
[17:25:55][29555][ERROR] Couldn't initialize CUDA driver API (error: 100)!
[17:25:55][29555][ERROR] Demodulation failed (error: 1020)!
17:25:55 (29555): called boinc_finish

            </stderr_txt>
            ]]>

            David Anderson
            David Anderson
            Joined: 3 Nov 10
            Posts: 14
            Credit: 123314944
            RAC: 0

            I turned off Einstein on this

            13 Aug 2016 2:08:38 UTC
            Message 148857
            (moderation:
              )

              I turned off Einstein on this particular computer.  it is getting too many errors

              on GPU tasks.  Worse than with the older NVidia driver.

               

               

               

               

              David Anderson
              David Anderson
              Joined: 3 Nov 10
              Posts: 14
              Credit: 123314944
              RAC: 0

              Well, this is embarrassing. 

              13 Aug 2016 18:43:29 UTC
              Message 148876
              (moderation:
                )

                Well, this is embarrassing.  I turned off Einstein on the wrong computer. Careless. So there is an error today, 13th, and another such is running and might error (letting it run to whatever completion it can manage).

                Corrected so the one with troublesome old machine will get no new Einstein tasks.

                The other two, with modern GPUs are doing Einstein.

                Sorry about the botch.

                 

                 

                Comment viewing options

                Select your preferred way to display the comments and click "Save settings" to activate your changes.

                Forums › Cruncher's Corner