Huge Workload

Forums › Problems and Bug Reports

illessa
illessa
Joined: 19 Aug 11
Posts: 5
Credit: 12479
RAC: 0
10 Oct 2011 15:17:04 UTC
Topic 196007
(moderation:
    )

    Hi,

    My latest download from Einstein@home shows a required working time of about 140 hrs! It's the Binary Radio Pulsar Search (Arecibo). Is this normal? I'll cancel this task as there is no chance to get it done in time...

    Brgds
    Sabine

    Richard Haselgrove
    Richard Haselgrove
    Joined: 10 Dec 05
    Posts: 2143
    Credit: 2991026312
    RAC: 703594

    Huge Workload

    10 Oct 2011 15:42:59 UTC
    Message 106879
    (moderation:
      )

      Even the server thought the task would take 3 million seconds, or 36 days!

      On the face of it, such a task shouldn't have been sent. Here's the server log - someone might have a look at it for you.

      Quote:
      2011-10-10 13:29:29.1420 [PID=27320] Request: [USER#xxxxx] [HOST#4197507] [IP xxx.xxx.xxx.103] client 6.12.34
      2011-10-10 13:29:29.1504 [PID=27320] [debug] [HOST#4197507] Resetting nresults_today
      2011-10-10 13:29:29.1505 [PID=27320] [send] effective_ncpus 2 max_jobs_on_host_cpu 999999 max_jobs_on_host 999999
      2011-10-10 13:29:29.1505 [PID=27320] [send] effective_ngpus 0 max_jobs_on_host_gpu 999999
      2011-10-10 13:29:29.1505 [PID=27320] [send] Not using matchmaker scheduling; Not using EDF sim
      2011-10-10 13:29:29.1505 [PID=27320] [send] CPU: req 1.00 sec, 1.00 instances; est delay 0.00
      2011-10-10 13:29:29.1506 [PID=27320] [send] work_req_seconds: 1.00 secs
      2011-10-10 13:29:29.1506 [PID=27320] [send] available disk 9.80 GB, work_buf_min 8640
      2011-10-10 13:29:29.1506 [PID=27320] [send] active_frac 0.783624 on_frac 0.207509 DCF 2.546825
      2011-10-10 13:29:29.1514 [PID=27320] [send] [HOST#4197507] is reliable
      2011-10-10 13:29:29.1515 [PID=27320] [send] set_trust: error rate 0.090250 > 0.050000, don't trust
      2011-10-10 13:29:29.1635 [PID=27320] [version] Checking plan class 'BRP3cuda32'
      2011-10-10 13:29:29.1636 [PID=27320] [version] Couldn't open plan class spec file '../plan_class_spec.xml'
      2011-10-10 13:29:29.1636 [PID=27320] [version] Host lacks CUDA coprocessor for plan class (BRP3cuda32)
      2011-10-10 13:29:29.1636 [PID=27320] [version] Checking plan class 'BRP3SSE'
      2011-10-10 13:29:29.1636 [PID=27320] [version] Couldn't open plan class spec file '../plan_class_spec.xml'
      2011-10-10 13:29:29.1636 [PID=27320] [version] Best version of app einsteinbinary_BRP4 is ID 288 (0.70 GFLOPS)
      2011-10-10 13:29:29.1653 [PID=27320] [debug] Sorted list of URLs follows [host timezone: UTC+7200]
      2011-10-10 13:29:29.1653 [PID=27320] [debug] zone=+03600 url=http://einstein-mirror.aei.uni-hannover.de/EatH
      2011-10-10 13:29:29.1653 [PID=27320] [debug] zone=-21600 url=http://einstein-dl4.phys.uwm.edu
      2011-10-10 13:29:29.1653 [PID=27320] [debug] zone=-21600 url=http://einstein-dl2.phys.uwm.edu
      2011-10-10 13:29:29.1653 [PID=27320] [debug] zone=-28800 url=http://einstein.ligo.caltech.edu
      2011-10-10 13:29:29.1655 [PID=27320] [send] [HOST#4197507] Sending app_version einsteinbinary_BRP4 2 100 BRP3SSE; 0.70 GFLOPS
      2011-10-10 13:29:29.1680 [PID=27320] [send] est. duration for WU 107108100: unscaled 199419.72 scaled 3123363.59
      2011-10-10 13:29:29.1680 [PID=27320] [HOST#4197507] Sending [RESULT#251769956 p2030.20100902.G55.34+01.16.S.b4s0g0.00000_1336_1] (est. dur. 3123363.59 seconds)
      2011-10-10 13:29:29.1687 [PID=27320] [send] don't need more work
      2011-10-10 13:29:29.1689 [PID=27320] [send] don't need more work
      2011-10-10 13:29:29.1689 [PID=27320] [send] don't need more work
      2011-10-10 13:29:29.1703 [PID=27320] Sending reply to [HOST#4197507]: 1 results, delay req 60.00
      2011-10-10 13:29:29.1707 [PID=27320] Scheduler ran 0.035 seconds
      illessa
      illessa
      Joined: 19 Aug 11
      Posts: 5
      Credit: 12479
      RAC: 0

      36 days? That's nice. I've

      10 Oct 2011 16:01:47 UTC
      Message 106880 in response to message 106879
      (moderation:
        )

        36 days? That's nice. I've received the task today and it will expire on Oct. 24th.

        Bikeman (Heinz-Bernd Eggenstein)
        Bikeman (Heinz-...
        Moderator
        Joined: 28 Aug 06
        Posts: 3522
        Credit: 779908412
        RAC: 1191986

        Sorry for that one. And

        11 Oct 2011 22:26:13 UTC
        Message 106881
        (moderation:
          )

          Sorry for that one. And thanks for reporting this problem, I notified the project staff about this.

          HBE

          illessa
          illessa
          Joined: 19 Aug 11
          Posts: 5
          Credit: 12479
          RAC: 0

          Hi folks, Einstein@home

          25 Dec 2011 13:46:43 UTC
          Message 106882
          (moderation:
            )

            Hi folks,

            Einstein@home starts to go crazy again. I've already cancelled 2 or 3 tasks because they were supposed to run for 130hrs. What's wrong here?

            Brgds
            Sabine

            Bikeman (Heinz-Bernd Eggenstein)
            Bikeman (Heinz-...
            Moderator
            Joined: 28 Aug 06
            Posts: 3522
            Credit: 779908412
            RAC: 1191986

            Hi, i'm afraid there is

            25 Dec 2011 18:28:27 UTC
            Message 106883 in response to message 106882
            (moderation:
              )

              Hi,

              i'm afraid there is nothing wrong with BOINC or Einstein@Home, but the host you are using has a rather slow Atom CPU. To put it in perspective, your Atom CPU is about as fast as a state of the art Pentium III CPU ... More than 10 years ago. The Atoms were designed to minimize power consumption, not for high performance. All the apps here at Einstein@Home make heavy use of floating point arithmetic, something the Atoms do not excel in.

              HBE

              illessa
              illessa
              Joined: 19 Aug 11
              Posts: 5
              Credit: 12479
              RAC: 0

              Understood. At least I know

              25 Dec 2011 18:49:17 UTC
              Message 106884
              (moderation:
                )

                Understood. At least I know now what's "wrong". In this case i continue to cancel task wich are too big. Thanks for info!

                Brgds
                Sabine

                Gary Roberts
                Gary Roberts
                Moderator
                Joined: 9 Feb 05
                Posts: 5877
                Credit: 118650213875
                RAC: 18832453

                RE: ... In this case i

                26 Dec 2011 13:06:37 UTC
                Message 106885 in response to message 106884
                (moderation:
                  )

                  Quote:
                  ... In this case i continue to cancel task wich are too big ...


                  In your previous message, you said, "Einstein@home starts to go crazy again." Actually it is you who is being a bit 'crazy' if you take the action you say you will :-). Let me suggest a better course of action.

                  If you study your computer's list of tasks, you will see two successfully completed tasks, a Gamma Ray Pulsar (FGRP) task and a Gravitational Wave (GW) task, both of which took close to 36 hours to complete. Whilst this is a fairly long time (because of the CPU you are using) it is well withing the deadline so there was no need for you to abort the two FGRP tasks that you did.

                  Your problem would appear to be the Binary Radio Pulsar (BRP) tasks using Arecibo data. These would be the tasks "supposed to run for 130hrs" that you mentioned. They would take a lot longer than FGRP or GW tasks (but perhaps not quite as long as 130 hours if you actually let one finish) so you should stop your computer from receiving them. To do that, you should go to your E@H preferences on the website and change the setting for 'Run only the selected applications'. All you need to do is click the 'edit' link and then untick the box next to 'Binary Radio Pulsar Search (Arecibo)'. Once you save your new preferences, you should click 'Update' in your BOINC Manager so that your BOINC client will know about the preference change immediately. Then you should get no further BRP tasks.

                  Cheers,
                  Gary.

                  Claggy
                  Claggy
                  Joined: 29 Dec 06
                  Posts: 560
                  Credit: 2751500
                  RAC: 1694

                  RE: To do that, you should

                  26 Dec 2011 14:36:00 UTC
                  Message 106886 in response to message 106885
                  (moderation:
                    )

                    Quote:
                    To do that, you should go to your E@H preferences on the website and change the setting for 'Run only the selected applications'. All you need to do is click the 'edit' link and then untick the box next to 'Binary Radio Pulsar Search (Arecibo)'. Once you save your new preferences, you should click 'Update' in your BOINC Manager so that your BOINC client will know about the preference change immediately. Then you should get no further BRP tasks.


                    If you're just changing the selected applications, Clicking update in Boinc Manager is Not required, this is a server side selection that isn't passed to the client,

                    Claggy

                    Jord
                    Joined: 26 Jan 05
                    Posts: 2952
                    Credit: 5893653
                    RAC: 1

                    Not for CPU tasks, no.

                    26 Dec 2011 14:59:01 UTC
                    Message 106887 in response to message 106886
                    (moderation:
                      )

                      Not for CPU tasks, no.

                      But you should for GPU tasks, or else you can find that BOINC will ask once for GPU work, and get as a message that no work for that will be downloaded.

                      Comment viewing options

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

                      Forums › Problems and Bug Reports