Do I have to do the whole thing myself?


Advanced search

Message boards : SZTAKI Desktop Grid : Do I have to do the whole thing myself?

AuthorMessage
Team Dockins
Send message
Joined: Nov 15 06
Posts: 2
Credit: 17,846
RAC: 0
Message 5434 - Posted 11 Jan 2007 18:42:21 UTC

    My latest SZTAKI \'segment\' has been running for over 3 days now and won\'t stop - 42 CPU hours and counting. The % completed stays frozen at 20% and the time to completion is counting UP! 51 MORE CPU hours to go and still increasing. NO OTHER BOINC project will run because this thing is hogging the whole BOINC project. What can I do to limit this beast??? Regards, Darryl Dockins dwdockins@charter.net

    robert.mouris
    Send message
    Joined: Nov 3 05
    Posts: 129
    Credit: 4,124,194
    RAC: 0
    Message 5435 - Posted 11 Jan 2007 19:20:21 UTC - in response to Message 5434.

      Last modified: 11 Jan 2007 19:27:10 UTC

      My latest SZTAKI \'segment\' has been running for over 3 days now and won\'t stop - 42 CPU hours and counting. The % completed stays frozen at 20% and the time to completion is counting UP! 51 MORE CPU hours to go and still increasing. NO OTHER BOINC project will run because this thing is hogging the whole BOINC project. What can I do to limit this beast??? Regards, Darryl Dockins dwdockins@charter.net

      Yes, you must do the whole thing yourself, a WU can\'t be shared by several crunchers. Unfortunately there is nothing you can do to limit the beast, unless you kill it and lose the 42 hours. Please don\'t bother because of the hours to go, as Boinc is not suited for this kind of project as regards the time to completion. Your WU might be finished in 1 hour (highly improbable), it might also need twice the 51 hours. If it still needs more, you had better kill the WU, because the program will crash around 200 hours CPU. Your computers are hidden, so we can\'t see what kind of WU it is, but I guess it is a 5-liner and that you\'re crunching the 2nd line. You can\'t deduce the crunching time needed by the next lines from that of the first lines.

      Please make sure that your project is set to \"no new work\". I forgot it once and got 99 such beasts in a few hours\' time! After the end of this WU, you must pay back your debt towards your other projects.
      ____________

      robert.mouris
      Send message
      Joined: Nov 3 05
      Posts: 129
      Credit: 4,124,194
      RAC: 0
      Message 5436 - Posted 11 Jan 2007 19:21:44 UTC - in response to Message 5435.

        Last modified: 11 Jan 2007 19:22:05 UTC


        ____________

        Team Dockins
        Send message
        Joined: Nov 15 06
        Posts: 2
        Credit: 17,846
        RAC: 0
        Message 5437 - Posted 11 Jan 2007 19:52:17 UTC

          Robert, thanks for your quick response. I think I will abort the task and suspend the application, or even perhaps detach from SZTAKI permanently as a lost cause. Is there any way to tell the server to only send me \'one-liners\'? If there is not, WHY does the BOINC system allow such an egregious resource hog to \'share\' its content? Also, why won\'t the thing let go and allow other projects equal time? I also notice that there is a mountain of complaints about no-credit issues. All in all, one poorly implemented program from a shared resource point of view, and it leaves me wondering whether even the completed tasks are providing the project with productive, trustworthy results. Thanks again - Darryl Dockins

          btw I am running a 1.25 GHZ eMac running OS X with plenty of RAM. I normally close off any viewing access simply because I am paranoid lest the typical peecee virus stuff gain access the my precious Mac. Am I too cautious here? Should I click \'yes\' to \'view my computer\' requests?

          larry1186
          Send message
          Joined: Sep 25 06
          Posts: 37
          Credit: 18,502
          RAC: 0
          Message 5438 - Posted 11 Jan 2007 21:12:45 UTC - in response to Message 5437.

            Last modified: 11 Jan 2007 21:14:36 UTC

            I normally close off any viewing access simply because I am paranoid lest the typical peecee virus stuff gain access the my precious Mac. Am I too cautious here? Should I click \'yes\' to \'view my computer\' requests?


            You don\'t have to worry one bit about others seeing your comps and doing bad things. All we can see are stats (owner, credit, type of CPU, OS, # results, etc) and no identifiable info is shared (IP, location could be narrowed down to country unless you specify \"international\" :) ). See here for what kind of info everybody else sees when \"viewing\" your comps. Cheers!

            [edit]edited for clarity[/edit]
            ____________
            Don't get distracted by shiny objects.

            robert.mouris
            Send message
            Joined: Nov 3 05
            Posts: 129
            Credit: 4,124,194
            RAC: 0
            Message 5442 - Posted 11 Jan 2007 22:26:29 UTC - in response to Message 5437.

              Is there any way to tell the server to only send me \'one-liners\'?

              There is no such way. The server checks your availability regarding RAM, CPU and project share and sends you what it estimates you can handle. For Sztaki the RAM requirements are very low and the CPU estimations don\'t work well (nowadays we have a lot of short WUs intermingled with recycled long ones, so the estimated average computing time is wrong in any case, and Boinc takes this when deciding how many WUs to send out).

              Because of the long WUs Boinc will have problems to finish the WUs in time, if you give a low percentage share to Sztaki. If you have several computers, it is better dedicating one mainly to Sztaki, rather than giving 5 or 10% to Sztaki on each computer.

              I also notice that there is a mountain of complaints about no-credit issues.

              Unfortunately, yes. Because some WUs are so long, and many crunchers are unable or unwilling to finish them and still knowing that they must wait many weeks to get a valid quorum, a lot of WUs must be reissued and this takes often 1 month. Sometimes there are too many WUs needed to get the quorum and the result is in error. And then credit is granted manually by the staff if there are too many WUs like that.

              I think I will abort the task and suspend the application, or even perhaps detach from SZTAKI permanently as a lost cause. All in all, one poorly implemented program from a shared resource point of view, and it leaves me wondering whether even the completed tasks are providing the project with productive, trustworthy results.

              According to the last posting from the project administrator, they are quite happy with the results. And they hope to get an easy-going project for us soon. If you want to be part of it, is of course your own decision. I don\'t want to lose a fellow cruncher, but if you can\'t be happy here, I wish you a lot of satisfaction with another project.

              Robert
              ____________

              Odysseus
              Avatar
              Send message
              Joined: Feb 27 06
              Posts: 212
              Credit: 221,397
              RAC: 0
              Message 5460 - Posted 13 Jan 2007 7:18:11 UTC - in response to Message 5442.

                Last modified: 13 Jan 2007 7:18:35 UTC

                Is there any way to tell the server to only send me \'one-liners\'?

                There is no such way.

                At any rate, IMO one-liners are the worst, because they spend the whole time showing 0% so you have no indication at all of the progress being made. My G4/400 server has been crunching one for 520 hours now, and I haven\'t the faintest idea how much longer it will take. At least the five-liners \'jump\' once in a while …
                ____________

                robert.mouris
                Send message
                Joined: Nov 3 05
                Posts: 129
                Credit: 4,124,194
                RAC: 0
                Message 5461 - Posted 13 Jan 2007 10:13:21 UTC - in response to Message 5460.

                  Last modified: 13 Jan 2007 10:16:06 UTC

                  My G4/400 server has been crunching one for 520 hours now

                  How on earth did you manage that? I see also that yesterday you finished one after 682 hours. Are the editors of the Guinness Book aware of your performance? Mine crash after 200+ hours, although I must say that I didn\'t get one of those for several months now.

                  ____________

                  Odysseus
                  Avatar
                  Send message
                  Joined: Feb 27 06
                  Posts: 212
                  Credit: 221,397
                  RAC: 0
                  Message 5462 - Posted 13 Jan 2007 10:21:24 UTC - in response to Message 5461.

                    My G4/400 server has been crunching one for 520 hours now

                    How on earth did you manage that? I see also that yesterday you finished one after 682 hours. Mine crash after 200+ hours, although I must say that I didn\'t get one of them for several months now.

                    Either you’re very lucky or I’m very unlucky, because most of the WUs I get seem to be tough ones …

                    I\'m mainly guessing, but I believe the “Maximum CPU time” is calculated from the benchmarks, being based on a notional number of FLOPs or clock-cycles. Slower machines would accordingly be allowed more time before triggering the error.
                    ____________

                    Profile Nightbird
                    Forum moderator
                    Avatar
                    Send message
                    Joined: Jul 12 05
                    Posts: 920
                    Credit: 114,924
                    RAC: 0
                    Message 5468 - Posted 14 Jan 2007 18:36:43 UTC

                      Last modified: 14 Jan 2007 18:38:08 UTC

                      A faster cpu would have a shorter maximum cpu time.
                      Look here please :
                      http://szdg.lpds.sztaki.hu/szdg/workunit.php?wuid=13772

                      computer 177674
                      Maximum CPU time exceeded after 484891 sec. (using the old application 2.02)
                      (and 28 * line 3)
                      ____________

                      Odysseus
                      Avatar
                      Send message
                      Joined: Feb 27 06
                      Posts: 212
                      Credit: 221,397
                      RAC: 0
                      Message 5479 - Posted 16 Jan 2007 10:29:55 UTC - in response to Message 5461.

                        […] I see also that yesterday you finished one after 682 hours. […]

                        BTW, I just noticed that the Pending Credit link on my Account page shows zero results, zero credits—yet my Results page shows four pending results claiming a total of about 2300 credits, of which that one WU accounts for over 2000.
                        ____________

                        Post to thread

                        Message boards : SZTAKI Desktop Grid : Do I have to do the whole thing myself?


                        Home | My Account | Message Boards


                        Copyright © 2017 SZTAKI Desktop Grid