Mac / R9 280c getting only computation errors

Forums › Problems and Bug Reports

Kailee71
Kailee71
Joined: 22 Nov 16
Posts: 35
Credit: 42623563
RAC: 0
2 Dec 2016 0:13:01 UTC
Topic 203344
(moderation:
    )

    Hi all,

    I just upgraded a Quadro K4000 to an R9 280x hoping to increase my output; however, none of the gpu tasks get beyond ca. 5 secs runtime; I just get computation errors. The only way I seem to be getting any gpu tasks at all is by enabling test applications. However, if I do enable these, I only get einsteinbinary_BRP4G_1.39_i686-apple-darwin__BRP4G-Beta-opencl-ati-Lion, where on the E@H website BRP4G-Beta-cuda55-Lion is listed as 1.56.

    What can I do to fix this/get my brand-new (used) R9 280x to do some work?

     

    Thanks a bunch in advance,

     

    Kailee71. 

    Richie
    Richie
    Joined: 7 Mar 14
    Posts: 656
    Credit: 1702989778
    RAC: 0

    Has your R9 280X been running

    2 Dec 2016 0:35:45 UTC
    Message 152334
    (moderation:
      )

      Has your R9 280X been running 1 task at a time or more in parallel ?

      Kailee71
      Kailee71
      Joined: 22 Nov 16
      Posts: 35
      Credit: 42623563
      RAC: 0

      Hi Richie,just one - not

      2 Dec 2016 7:42:36 UTC
      Message 152338
      (moderation:
        )

        Hi Richie,

        just one - not even that I can get running... :-/

        One thing that might be relevant; as I said, I upgraded from a Quadro 4000, and had to uninstall the Mac Cuda Drivers manually. Maybe someone could list the files to be deleted just in case I left overlooked something/

         Also; here's my cc_config.xml:

        <cc_config>

        <options>

        <use_all_gpus>1</use_all_gpus>

        <ncpus>32</ncpus> 

        <report_results_immediately>1</report_results_immediately>

        <allow_remote_gui_rpc>1</allow_remote_gui_rpc>

        </options>

        </cc_config>

        I'm running without an app_config.xml. I just tried resetting the project, and again got this;
         

        Fri 2 Dec 08:14:08 2016 | Einstein@Home | Started download of einsteinbinary_BRP4G_1.39_i686-apple-darwin__BRP4G-Beta-opencl-ati-Lion

        .

        .

        .

        Fri 2 Dec 08:14:32 2016 | Einstein@Home | Output file p2030.20160204.G178.39-01.35.S.b0s0g0.00000_16_0_0 for task p2030.20160204.G178.39-01.35.S.b0s0g0.00000_16_0 absent

        Fri 2 Dec 08:14:32 2016 | Einstein@Home | Output file p2030.20160204.G178.39-01.35.S.b0s0g0.00000_16_0_1 for task p2030.20160204.G178.39-01.35.S.b0s0g0.00000_16_0 absent
        Fri 2 Dec 08:14:32 2016 | Einstein@Home | Output file p2030.20160204.G178.39-01.35.S.b0s0g0.00000_16_0_2 for task p2030.20160204.G178.39-01.35.S.b0s0g0.00000_16_0 absent
        Fri 2 Dec 08:14:32 2016 | Einstein@Home | Output file p2030.20160204.G178.39-01.35.S.b0s0g0.00000_16_0_3 for task p2030.20160204.G178.39-01.35.S.b0s0g0.00000_16_0 absent
        Fri 2 Dec 08:14:32 2016 | Einstein@Home | Output file p2030.20160204.G178.39-01.35.S.b0s0g0.00000_16_0_4 for task p2030.20160204.G178.39-01.35.S.b0s0g0.00000_16_0 absent
        Fri 2 Dec 08:14:32 2016 | Einstein@Home | Output file p2030.20160204.G178.39-01.35.S.b0s0g0.00000_16_0_5 for task p2030.20160204.G178.39-01.35.S.b0s0g0.00000_16_0 absent
        Fri 2 Dec 08:14:32 2016 | Einstein@Home | Output file p2030.20160204.G178.39-01.35.S.b0s0g0.00000_16_0_6 for task p2030.20160204.G178.39-01.35.S.b0s0g0.00000_16_0 absent
        Fri 2 Dec 08:14:32 2016 | Einstein@Home | Output file p2030.20160204.G178.39-01.35.S.b0s0g0.00000_16_0_7 for task p2030.20160204.G178.39-01.35.S.b0s0g0.00000_16_0 absent
        Fri 2 Dec 08:14:32 2016 | Einstein@Home | Output file p2030.20160204.G178.39-01.35.S.b0s0g0.00000_16_0_8 for task p2030.20160204.G178.39-01.35.S.b0s0g0.00000_16_0 absent
        Fri 2 Dec 08:14:32 2016 | Einstein@Home | Output file p2030.20160204.G178.39-01.35.S.b0s0g0.00000_16_0_9 for task p2030.20160204.G178.39-01.35.S.b0s0g0.00000_16_0 absent
        Fri 2 Dec 08:14:32 2016 | Einstein@Home | Output file p2030.20160204.G178.39-01.35.S.b0s0g0.00000_16_0_10 for task p2030.20160204.G178.39-01.35.S.b0s0g0.00000_16_0 absent
        Fri 2 Dec 08:14:32 2016 | Einstein@Home | Output file p2030.20160204.G178.39-01.35.S.b0s0g0.00000_16_0_11 for task p2030.20160204.G178.39-01.35.S.b0s0g0.00000_16_0 absent
        Fri 2 Dec 08:14:32 2016 | Einstein@Home | Output file p2030.20160204.G178.39-01.35.S.b0s0g0.00000_16_0_12 for task p2030.20160204.G178.39-01.35.S.b0s0g0.00000_16_0 absent
        Fri 2 Dec 08:14:32 2016 | Einstein@Home | Output file p2030.20160204.G178.39-01.35.S.b0s0g0.00000_16_0_13 for task p2030.20160204.G178.39-01.35.S.b0s0g0.00000_16_0 absent
        Fri 2 Dec 08:14:32 2016 | Einstein@Home | Output file p2030.20160204.G178.39-01.35.S.b0s0g0.00000_16_0_14 for task p2030.20160204.G178.39-01.35.S.b0s0g0.00000_16_0 absent
        Fri 2 Dec 08:14:32 2016 | Einstein@Home | Output file p2030.20160204.G178.39-01.35.S.b0s0g0.00000_16_0_15 for task p2030.20160204.G178.39-01.35.S.b0s0g0.00000_16_0 absent

        And here's some stderr from one of the tasks;

        https://einsteinathome.org/task/592308624

        Shouldn't I be getting the 1.56 BRP4G as per https://einsteinathome.org/apps.php? And what's with the absent output files?

        Tia,

         

        Kailee.

        Richie
        Richie
        Joined: 7 Mar 14
        Posts: 656
        Credit: 1702989778
        RAC: 0

        Kai Leibrandt wrote:Shouldn't

        2 Dec 2016 13:02:39 UTC
        Message 152346 in response to message 152338
        (moderation:
          )

          Kai Leibrandt wrote:
          Shouldn't I be getting the 1.56 BRP4G as per https://einsteinathome.org/apps.php? And what's with the absent output files?

          Hi Tia,

          1.56 is a cuda-application... 1.56 (BRP4G-Beta-cuda55-Lion) (beta test). I don't know anything about MACs and wasn't sure if they could perform some kind of cuda-simulation with ATI/AMD cards. I googled "BRP4G-Beta-cuda55-Lion" and looked at the specs of those computers that have produced valid results today. They all have Nvidia cards.

          I believe your computer would need a genuine Nvidia card to receive those 1.56 tasks.

          Holmis
          Joined: 4 Jan 05
          Posts: 1118
          Credit: 1055935564
          RAC: 0

          Not running a Mac but I would

          2 Dec 2016 13:46:34 UTC
          Message 152347
          (moderation:
            )

            Not running a Mac but I would start by trying a different driver for the R9 280x, preferably newer but if the newest is already installed then an older one. Make sure the driver includes OpenCL support.

            You want the OpenCL app as that's the one developed for AMD cards, the CUDA app is for Nvidia cards.

            Boinc expects the app to produce a number of output files, when the app crashes those files can't be created so Boinc complains about them not being there.

            Kailee71
            Kailee71
            Joined: 22 Nov 16
            Posts: 35
            Credit: 42623563
            RAC: 0

            Doh! Should have seen that

            2 Dec 2016 22:37:38 UTC
            Message 152348
            (moderation:
              )

              Doh! Should have seen that 1.56 is cuda... Sorry.

               

              Will try to find drivers from AMD (using stock Apple drivers at the mo), and report back.

               

              Thanks!

               

              Kai.

               

               

              EDIT: No drivers available separately from AMD that I can find... So - anything else I can do?

              Comment viewing options

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

              Forums › Problems and Bug Reports