Very strange!!!

Forums › Cruncher's Corner

GreyCruncher
GreyCruncher
Joined: 2 Sep 06
Posts: 22
Credit: 28664453
RAC: 0
12 Sep 2011 19:36:30 UTC
Topic 195963
(moderation:
    )

    http://einsteinathome.org/workunit/104824958

    Watch run time and CPU time. nothing was changed in settings.

    Betreger
    Betreger
    Joined: 25 Feb 05
    Posts: 992
    Credit: 1583945796
    RAC: 756060

    Very strange!!!

    12 Sep 2011 20:10:53 UTC
    Message 106649
    (moderation:
      )

      Who knows, this may be useful in time travel.

      GreyCruncher
      GreyCruncher
      Joined: 2 Sep 06
      Posts: 22
      Credit: 28664453
      RAC: 0

      Always take a towel with you.

      12 Sep 2011 20:19:37 UTC
      Message 106650 in response to message 106649
      (moderation:
        )

        Always take a towel with you. *chrchr*

        Gary Roberts
        Gary Roberts
        Moderator
        Joined: 9 Feb 05
        Posts: 5872
        Credit: 117331664762
        RAC: 35783372

        RE: http://einstein.phys.uw

        13 Sep 2011 1:52:36 UTC
        Message 106651
        (moderation:
          )

          Quote:

          http://einsteinathome.org/workunit/104824958

          Watch run time and CPU time. nothing was changed in settings.


          Actually not very strange at all :-).

          You need to click on the taskID for your task and have a read through the stderr output text that was returned to the project. Here is the relevant bit right down near the end of the output. Note that there is a limit on what is returned so you can't see the entire file. Approximately the first half is missing and the file starts just as checkpoint 23 (out of 50) was being written.

          Quote:

          .....
          % Time spent on semicoherent stage: 578 s
          % Starting coherent stage.
          % Time spent on coherent stage: 137 s
          % checkpoint written: skypoint 49
          % Path to ephemeris file: ../../projects/einstein.phys.uwm.edu/JPLEPH
          % Starting barycentering for sky point 50 / 50
          % Creating FFT plan.
          % Starting semicoherent stage.
          % Status 1/83 ( 5 s)
          % Status 2/83 ( 5 s)
          % Status 3/83 ( 5 s)
          % Status 4/83 ( 5 s)
          % Status 5/83 ( 5 s)
          % Status 6/83 ( 5 s)
          % Status 7/83 ( 5 s)
          % Status 8/83 ( 5 s)
          % Status 9/83 ( 5 s)
          % Status 10/83 ( 5 s)
          % Status 11/83 ( 5 s)
          % Status 12/83 ( 5 s)
          % Status 13/83 ( 5 s)
          % Status 14/83 ( 4 s)
          % Status 15/83 ( 4 s)
          % Status 16/83 ( 4 s)
          % Status 17/83 ( 4 s)
          % Status 18/83 ( 4 s)
          % Status 19/83 ( 5 s)
          % Status 20/83 ( 5 s)
          % Status 21/83 ( 5 s)
          % Status 22/83 ( 5 s)
          % Status 23/83 ( 5 s)
          % Status 24/83 ( 6 s)
          % Status 25/83 ( 5 s)
          % Status 26/83 ( 5 s)
          % Status 27/83 ( 6 s)
          % Status 28/83 ( 5 s)
          % Status 29/83 ( 5 s)
          % Status 30/83 ( 6 s)
          % Status 31/83 ( 5 s)
          % Status 32/83 ( 5 s)
          % Status 33/83 ( 5 s)
          07:30:39 (2895): [normal]: This Einstein@home App was built at: Aug 30 2011 14:23:52

          07:30:39 (2895): [normal]: Start of BOINC application '../../projects/einstein.phys.uwm.edu/hsgamma_FGRP1_0.23_i686-pc-linux-gnu'.
          command line: ../../projects/einstein.phys.uwm.edu/hsgamma_FGRP1_0.23_i686-pc-linux-gnu --inputfile ../../projects/einstein.phys.uwm.edu/LATeah0024T.dat --outputfile results.cand.out --f0start 32.0 --f0Band 64.0 --BskyPointFile ../../projects/einstein.phys.uwm.edu/skygrid_LATeah0024T_0096.0.dat --firstSkyPoint 0 --numSkyPoints 50 --f1dot -1.7e-11 --f1dotBand 1e-12 --ephemdir ../../projects/einstein.phys.uwm.edu/JPLEPH --Tcoh 524288.0 --toplist 10 --useWeights 1 --cohFollow 4 --inCohTimeRatio 35 --debug 1
          output files: 'results.cand.out' '../../projects/einstein.phys.uwm.edu/LATeah0024T_96.0_0_-1.6e-11_0_0' 'results.cand.out.cohfu' '../../projects/einstein.phys.uwm.edu/LATeah0024T_96.0_0_-1.6e-11_0_1'
          07:30:39 (2895): [debug]: Flags: i386 SSE GNUC X86 GNUX86
          07:30:39 (2895): [debug]: glibc version/release: 2.13/stable
          07:30:39 (2895): [debug]: Set up communication with graphics process.
          % Opening inputfile: ../../projects/einstein.phys.uwm.edu/LATeah0024T.dat
          % Total amount of photon times: 8000
          % Preparing toplist of length: 10
          % Read 50 sky positions
          % checkpoint read: skypoint 49


          % Path to ephemeris file: ../../projects/einstein.phys.uwm.edu/JPLEPH
          % Starting barycentering for sky point 50 / 50
          % Creating FFT plan.
          % Starting semicoherent stage.
          % Status 1/83 ( 6 s)
          % Status 2/83 ( 5 s)
          % Status 3/83 ( 5 s)
          % Status 4/83 ( 5 s)
          ....


          I've made bold a couple of key bits. The first bit shows the writing of the 49th checkpoint (out of 50) and the starting of work on checkpoint 50. Processing is almost done (98% complete) with just one stage to go. For some reason before completion of that final stage, the app was restarted and checkpoint 49 was read in from disk. See the second bold bit. The run time was just 750 secs for this last 2% and that's what got recorded instead of the total run time. It seems that restarting the app causes the original runtime to be discarded or reset to zero for some reason. I'm sure I've seen this behaviour before - perhaps it's something to do with Linux. If one checkpoint takes 750 secs you can work out an estimate of the true runtime - something like 50x750=37.500 secs which fits pretty well with the recorded CPU time.

          If you think there is something funny with a particular task, it's always useful to browse what is returned to the project to see if there is relevant information there.

          Cheers,
          Gary.

          Comment viewing options

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

          Forums › Cruncher's Corner