New application W2.03/W2.04 - Feedback


Advanced search

Message boards : SZTAKI Desktop Grid : New application W2.03/W2.04 - Feedback

AuthorMessage
Stick
Send message
Joined: Jun 12 06
Posts: 193
Credit: 66,244
RAC: 0
Message 4721 - Posted 21 Nov 2006 14:53:37 UTC

    I just noticed that the \"News\" on the Home Page has an item, dated 2006 Nov 20, regarding W2.02. Is this the same 2.02 that was released on Nov 6 or is it another one? (Normally, that would be considered a \"stupid question\" but since there were 2 different 2.02\'s out there already, the idea of a third isn\'t that farfetched.)
    ____________

    larry1186
    Send message
    Joined: Sep 25 06
    Posts: 37
    Credit: 18,502
    RAC: 0
    Message 4722 - Posted 21 Nov 2006 15:25:55 UTC

      Last modified: 21 Nov 2006 15:39:44 UTC

      Are we going back in time?
      2006-11-20 21:51:47 [SZTAKI Desktop Grid] Started download of file search_1.00_windows_intelx86.exe
      2006-11-20 21:51:51 [SZTAKI Desktop Grid] Finished download of file search_1.00_windows_intelx86.exe

      The stderr.txt says it used search 1.00 but it was reported as search 2.03...?? At least the WUs are finishing (WUs 31819, 31857, and 8952 finished so far and a fourth at 34% after 2 hrs).

      [edit]BOINC manager task pane, application column says search 2.03[/edit]

      [edit #2]from client_stat.xml:
      [code]
      -app_version-
      -app_name- search -/app_name-
      -version_num- 203 -/version_num-
      -file_ref-
      -file_name- search_1.00_windows_intelx86.exe -/file_name-
      -main_program/-
      -/file_ref-
      ...
      -/app_version-
      (BTW, I couldn\'t get the [ code ] tag to work so I replaced \'<\' \'>\' with \'-\')
      so it\'s 1.00 looking like 2.03 in every way.[/edit #2]

      ____________
      Don't get distracted by shiny objects.

      Stick
      Send message
      Joined: Jun 12 06
      Posts: 193
      Credit: 66,244
      RAC: 0
      Message 4723 - Posted 21 Nov 2006 15:55:50 UTC - in response to Message 4721.

        Last modified: 21 Nov 2006 16:10:16 UTC

        I just noticed that the \"News\" on the Home Page has an item, dated 2006 Nov 20, regarding W2.02. Is this the same 2.02 that was released on Nov 6 or is it another one? (Normally, that would be considered a \"stupid question\" but since there were 2 different 2.02\'s out there already, the idea of a third isn\'t that farfetched.)


        I guess the \"News\" item has a typo. It appears that W2.03 is the new version. (But, I now have a \"Search 1.00\" app in my projects folder, too - even though it also says 2.03 under the Tasks tab.)
        ____________

        Profile Nightbird
        Forum moderator
        Avatar
        Send message
        Joined: Jul 12 05
        Posts: 920
        Credit: 114,924
        RAC: 0
        Message 4728 - Posted 21 Nov 2006 20:22:05 UTC

          Last modified: 24 Nov 2006 19:28:19 UTC

          new thread
          some messages moved ;)
          ____________

          Stick
          Send message
          Joined: Jun 12 06
          Posts: 193
          Credit: 66,244
          RAC: 0
          Message 4741 - Posted 23 Nov 2006 1:03:59 UTC

            Result ID 166546 is my first WU using W2.03 on my Intel(R) Pentium(R) M processor 1.80GHz with WinXP Pro and BOINC 5.4.11 (and it finished OK).
            ____________

            ifg@work
            Send message
            Joined: Jul 12 05
            Posts: 13
            Credit: 103,819
            RAC: 0
            Message 4766 - Posted 24 Nov 2006 12:01:42 UTC

              Hmmm,
              forced my computer to download new WUs and received W2.04 instead of W2.03.

              BTW.: I\'m running a Pentium M @ 1,6 GHz, WinXP Pro, 1GB, and BOINC 5.7.5 (in the moment).

              GRS
              Send message
              Joined: Oct 3 06
              Posts: 1
              Credit: 154,453
              RAC: 0
              Message 4807 - Posted 26 Nov 2006 17:42:37 UTC

                First work OK, second slower, but next one takes 10 hours. If you restart project - everything statrs again.
                Is it normal? ;)

                ____________

                akosf
                Avatar
                Send message
                Joined: Aug 30 05
                Posts: 62
                Credit: 510,419
                RAC: 0
                Message 5005 - Posted 11 Dec 2006 16:23:45 UTC

                  The W2.04 application has a checkpoint problem. It skips over the remainder part of the stopped line when the application is restarted with a half-processed WU.

                  This mechanism is usable to eliminate the long WUs on the user side.
                  (eg.: 2 procejts with small switching time)

                  Jack Narcotta
                  Avatar
                  Send message
                  Joined: Jul 15 05
                  Posts: 20
                  Credit: 5,924
                  RAC: 0
                  Message 5159 - Posted 19 Dec 2006 15:53:03 UTC

                    2.04 Running fine on this machine, last few results have been stable. Much better.
                    ____________

                    Gaurav
                    Send message
                    Joined: Sep 4 06
                    Posts: 22
                    Credit: 52,275
                    RAC: 0
                    Message 5258 - Posted 21 Dec 2006 6:39:43 UTC - in response to Message 5159.

                      Last modified: 21 Dec 2006 6:40:53 UTC

                      on my machine 2.04 ver. shows 20% of a wu completed in 10 min. but the wu is completed in 20hrs or more. currently a wu running on ver 2.04 showed 20% completion in the first 11min. but it is 60% completed now in 34hours.
                      my computer is
                      Pentium D 2.8ghz
                      512MB ram
                      windows Xp pro SP2
                      running boinc 5.4.11


                      I don\'t know if this is normal or not. I\'v had to abort 4 or 5 wu\'s in the past because in about 30hrs each had completed only 20% which it had shown in the first 10 min.

                      ____________

                      Odysseus
                      Avatar
                      Send message
                      Joined: Feb 27 06
                      Posts: 212
                      Credit: 221,397
                      RAC: 0
                      Message 5277 - Posted 22 Dec 2006 6:04:19 UTC - in response to Message 5258.

                        on my machine 2.04 ver. shows 20% of a wu completed in 10 min. but the wu is completed in 20hrs or more. currently a wu running on ver 2.04 showed 20% completion in the first 11min. but it is 60% completed now in 34hours.

                        SGD WUs contain a number of lines, each of which can take a completely different amount of time to process than the others do. The latter WU you describe probably has five lines, the first taking only 11 minutes and the next two 34 hours between them (which could be 17 h each, or 10 min. & 33h50m, or anything in between…).

                        BOINC’s completion-time estimates don’t handle this situation well; you shouldn’t base any decisions on those figures.
                        ____________

                        hans van der giessen
                        Send message
                        Joined: Mar 11 06
                        Posts: 2
                        Credit: 14,674
                        RAC: 0
                        Message 5284 - Posted 23 Dec 2006 7:49:00 UTC

                          I suspended participation in Sztaki desktop grid (version 2.04) today. Reasons:

                          - \'To completion\' and \'Progress\' indicators in the BOINC client are not correct (completion time increases);
                          - because I do not understand Hungarian it is relatively hard (compared to other BOINC projects) to get information.

                          Regards,

                          Hans van der Giessen
                          Netherlands

                          ____________

                          robert.mouris
                          Send message
                          Joined: Nov 3 05
                          Posts: 129
                          Credit: 4,124,189
                          RAC: 0
                          Message 5301 - Posted 26 Dec 2006 12:50:19 UTC - in response to Message 5284.

                            Last modified: 26 Dec 2006 12:51:28 UTC

                            \'To completion\' and \'Progress\' indicators in the BOINC client are not correct (completion time increases);

                            The \"progress\" indicator is correct as far as the number of processed lines is concerned. But it doesn\'t increase regularly with elapsed time. When starting to process a line, the program cannot know the required time. The \"to completion\" indicator is wrong and does indeed increase during the processing of a line. Only at the end of the processing of each line, it decreases sharply, and then increases again. This is annoying, but is a Boinc feature, which doesn\'t fit well to the particular needs of this project. My solution: just ignore it. If, however, you need a project with highly regular and foreseeable WUs, then I\'m afraid you will not be happy here and had better choose another project. Sorry if you left, but this is the nature of this project...
                            because I do not understand Hungarian it is relatively hard (compared to other BOINC projects) to get information.

                            The whole information is available in English, except one thread on this board where participants converse in Hungarian. If you want to read the FAQ in English, you must first click on the British flag on the homepage. This is a bug in the system.

                            ____________

                            hans van der giessen
                            Send message
                            Joined: Mar 11 06
                            Posts: 2
                            Credit: 14,674
                            RAC: 0
                            Message 5319 - Posted 27 Dec 2006 8:04:28 UTC - in response to Message 5301.

                              Robert, thanks for your reply,

                              \'To completion\' and \'Progress\' indicators in the BOINC client are not correct (completion time increases);


                              <cut>My solution: just ignore it. If, however, you need a project with highly regular and foreseeable WUs, then I\'m afraid you will not be happy here and had better choose another project. Sorry if you left, but this is the nature of this project...


                              My BOINC clients are attached to a lot of projects. Some projects do not finish in time because the Sztaki project WU\'s need more time while they are running (the time to completion even increases).

                              because I do not understand Hungarian it is relatively hard (compared to other BOINC projects) to get information.


                              The whole information is available in English, except one thread on this board where participants converse in Hungarian. If you want to read the FAQ in English, you must first click on the British flag on the homepage. This is a bug in the system.


                              The information may be available in English (yes, I clicked the British flag from the start) but most, if not all, of the controls are in Hungarian. For instance, I clicked \"Válasz\" to reply to your message within this \"Üzenőtáblák\". That makes the information, like I wrote, relatively hard to access.

                              Regards,

                              Hans van der Giessen
                              Netherlands
                              ____________

                              robert.mouris
                              Send message
                              Joined: Nov 3 05
                              Posts: 129
                              Credit: 4,124,189
                              RAC: 0
                              Message 5320 - Posted 27 Dec 2006 8:43:06 UTC - in response to Message 5319.

                                Last modified: 27 Dec 2006 8:51:23 UTC

                                My BOINC clients are attached to a lot of projects.

                                This unpredictable length of WUs is indeed a problem to Boinc. May I suggest 2 ways to make it easier:

                                • You have attached 2 computers to Sztaki. You might attach just 1 and double its share.
                                • I see that you are also attached to Climateprediction. You might suspend this, give full power to Sztaki for the time it needs, and after that suspend Sztaki and \"repay\" your debt to Climateprediction; you must of course select \"no new task\", otherwise it can\'t work. Climateprediction has long deadlines and won\'t see a difference. After some time you switch back to Sztaki for a while.


                                The information may be available in English (yes, I clicked the British flag from the start) but most, if not all, of the controls are in Hungarian. For instance, I clicked \"Válasz\" to reply to your message within this \"Üzenőtáblák\". That makes the information, like I wrote, relatively hard to access.

                                This is strange, because I just clicked on \"Reply to this post\" inside this \"Message\". I don\'t know what I did, but I have everywhere English right from the start (as far as it was translated in the beginning). Maybe someone else can tell us?

                                Robert
                                ____________

                                gintaras46
                                Send message
                                Joined: Feb 17 06
                                Posts: 3
                                Credit: 19,315
                                RAC: 0
                                Message 5325 - Posted 27 Dec 2006 22:36:18 UTC

                                  On all of my PC\'s running Sztaki, there are 20% with 24 - 50 Hours of time for crunching spent. I\'ve quited most of other projects WU\'s, just to be in time with Sztaki, but now I\'m quitting Sztaki for another half a year. It takes too long to crunch theyr WU\'s. There is no need to equalize Sztaki with Climateprediction.net, because climateprediction is sending the \"trickle up\" updates from time to time and for those we are getting credits. Sztaki does not. Even worse, when for, let\'s say 60000 seconds of work you are getting 0.00 credits.

                                  Will be back after problems are solved.

                                  Gintaras

                                  Alex Plantema
                                  Send message
                                  Joined: Jul 16 05
                                  Posts: 10
                                  Credit: 5,221
                                  RAC: 0
                                  Message 5383 - Posted 2 Jan 2007 19:57:01 UTC

                                    I\'ve had several work units with version 2.04 with running times between 22 minutes and 26 hours, but now I have a work unit already running 40 hours and still at 0%. Should I abort it?
                                    ____________

                                    Profile jamin
                                    Avatar
                                    Send message
                                    Joined: Nov 20 06
                                    Posts: 324
                                    Credit: 100,110
                                    RAC: 0
                                    Message 5384 - Posted 2 Jan 2007 20:05:09 UTC

                                      Hard to tell - you supposedly have a 1liner - it won\'t show progress until it\'s finished :)
                                      If you\'re planning to restart BOINC or your computer in following 40hours it will fail with \"empty output\"

                                      It could be good to check if there are chances for you to get credits for this WU by looking how others managed with this particular WU :)
                                      ____________
                                      Bang! Zoom! Right in the kisser!

                                      robert.mouris
                                      Send message
                                      Joined: Nov 3 05
                                      Posts: 129
                                      Credit: 4,124,189
                                      RAC: 0
                                      Message 5385 - Posted 2 Jan 2007 20:25:49 UTC

                                        Last modified: 2 Jan 2007 20:30:23 UTC

                                        I suppose that it is this one? Yes, it is a 1-line WU.

                                        You must be ready to wait several weeks to get credit because the other 3 crunchers have restarted the WU and got the \"empty output\" message\". So it must still be reissued 2 times after you have finished.
                                        ____________

                                        Profile nubz
                                        Send message
                                        Joined: Dec 25 05
                                        Posts: 6
                                        Credit: 236,204
                                        RAC: 0
                                        Message 5389 - Posted 4 Jan 2007 14:11:44 UTC

                                          Can you check http://szdg.lpds.sztaki.hu/szdg/workunit.php?wuid=18775 - it\'s been pending for about a week. Is that normal?
                                          ____________

                                          larry1186
                                          Send message
                                          Joined: Sep 25 06
                                          Posts: 37
                                          Credit: 18,502
                                          RAC: 0
                                          Message 5390 - Posted 4 Jan 2007 15:31:26 UTC - in response to Message 5389.

                                            Can you check http://szdg.lpds.sztaki.hu/szdg/workunit.php?wuid=18775 - it\'s been pending for about a week. Is that normal?


                                            Well, normal as far as one-line WUs are concerned. All three successfully returned results are empty - so in order for the WU to validate, three non-empty results will have to be returned which will take a while since it will be distributed one at a time. This isn\'t specific to W2.03/W2.04, it has something to do with the checkpointing and it\'s been around for a while (at least in the 2.x versions that I\'ve used).
                                            ____________
                                            Don't get distracted by shiny objects.

                                            Profile Bruno G. Olsen & ESEA @ greenholt
                                            Avatar
                                            Send message
                                            Joined: Jul 12 05
                                            Posts: 16
                                            Credit: 19,160
                                            RAC: 0
                                            Message 5397 - Posted 6 Jan 2007 5:02:26 UTC

                                              maybe estimated time to completion set on server should be increased? The bunch of wu\'s I got is estimated to have a duration of 2 hours and twenty minutes, so that is the time used to figure out how much I should download - not good if it turns out they\'re all 40 hour wu\'s ;)
                                              ____________

                                              Profile Narwhal
                                              Avatar
                                              Send message
                                              Joined: Aug 25 05
                                              Posts: 32
                                              Credit: 688,804
                                              RAC: 0
                                              Message 5400 - Posted 6 Jan 2007 16:18:24 UTC - in response to Message 5389.

                                                Can you check http://szdg.lpds.sztaki.hu/szdg/workunit.php?wuid=18775 - it\'s been pending for about a week. Is that normal?


                                                Only a week this one been pending since 24 Nov 2006
                                                http://szdg.lpds.sztaki.hu/szdg/result.php?resultid=172906
                                                ____________

                                                Alex Plantema
                                                Send message
                                                Joined: Jul 16 05
                                                Posts: 10
                                                Credit: 5,221
                                                RAC: 0
                                                Message 5405 - Posted 7 Jan 2007 9:46:17 UTC - in response to Message 5385.

                                                  I suppose that it is this one? Yes, it is a 1-line WU.

                                                  You must be ready to wait several weeks to get credit because the other 3 crunchers have restarted the WU and got the \"empty output\" message\". So it must still be reissued 2 times after you have finished.

                                                  That\'s the one. It finished after 107 hours, longer than any other workunit I could find (the longest with version 2.04 was 61 hours). I wonder if two more users will have the same patience. I\'ve got more workunits pending with shorter running times. Of the 1-liners I processed none have gotten credit granted so far.

                                                  If you\'re planning to restart BOINC or your computer in following 40hours it will fail with \"empty output\"

                                                  It looks like there\'s no decent checkpointing.
                                                  ____________

                                                  Stick
                                                  Send message
                                                  Joined: Jun 12 06
                                                  Posts: 193
                                                  Credit: 66,244
                                                  RAC: 0
                                                  Message 5406 - Posted 7 Jan 2007 14:37:13 UTC - in response to Message 5405.

                                                    Look at wuid=626. It\'s a 21b342bd series WU and it validated yesterday - so it is not impossible. But, it is the first one to do so that I have seen. In fact, Result ID 294727 was the first 21b342bd series result I had ever seen that did NOT end with an \"Output is empty\" message. And that was the reason, I decided not to abort Result ID 330515 when I got it. But (as I have said before), being able to go 30+ hours (and in your case 100+ hours) without restarting your computer, is a tall order.

                                                    I suppose that it is this one? Yes, it is a 1-line WU.

                                                    You must be ready to wait several weeks to get credit because the other 3 crunchers have restarted the WU and got the \"empty output\" message\". So it must still be reissued 2 times after you have finished.

                                                    That\'s the one. It finished after 107 hours, longer than any other workunit I could find (the longest with version 2.04 was 61 hours). I wonder if two more users will have the same patience. I\'ve got more workunits pending with shorter running times. Of the 1-liners I processed none have gotten credit granted so far.

                                                    If you\'re planning to restart BOINC or your computer in following 40hours it will fail with \"empty output\"

                                                    It looks like there\'s no decent checkpointing.


                                                    ____________

                                                    Alex Plantema
                                                    Send message
                                                    Joined: Jul 16 05
                                                    Posts: 10
                                                    Credit: 5,221
                                                    RAC: 0
                                                    Message 5408 - Posted 7 Jan 2007 15:36:33 UTC - in response to Message 5406.

                                                      Look at wuid=626. It\'s a 21b342bd series WU and it validated yesterday - so it is not impossible. But, it is the first one to do so that I have seen. In fact, Result ID 294727 was the first 21b342bd series result I had ever seen that did NOT end with an \"Output is empty\" message. And that was the reason, I decided not to abort Result ID 330515 when I got it.

                                                      It\'s strange that though the minimum quorum of wuid=626 is 3, credit has been granted after only 2 valid results. Some results were were obtained using version 2.00.

                                                      ____________

                                                      Stick
                                                      Send message
                                                      Joined: Jun 12 06
                                                      Posts: 193
                                                      Credit: 66,244
                                                      RAC: 0
                                                      Message 5410 - Posted 7 Jan 2007 17:30:14 UTC - in response to Message 5408.

                                                        It\'s strange that though the minimum quorum of wuid=626 is 3, credit has been granted after only 2 valid results.


                                                        I had been confused about quorums, too. This thread does a pretty good job of explaining it.

                                                        Some results were were obtained using version 2.00.


                                                        Yeah, but the real problem is the checkpointing/empty output issue which is still there (in v2.04). It\'s just that 2.04 is faster (and, therefore finishes up some WU\'s before people notice that they are \"stuck\"). If you\'d like to read a little more about the checkpointing problem, these threads have some good info: here and here.

                                                        ____________

                                                        Odysseus
                                                        Avatar
                                                        Send message
                                                        Joined: Feb 27 06
                                                        Posts: 212
                                                        Credit: 221,397
                                                        RAC: 0
                                                        Message 5422 - Posted 8 Jan 2007 22:15:09 UTC

                                                          My Win98 system finally completed a WU with v2.04 without erroring out, after nearly 140 hours … but the result was designated as invalid. This means that over the last three months it’s contributed nothing except 600 hours of time that could have been useful to another project. Needless to say, I’m rather disappointed; if the next result is a failure—of whatever kind—I’ll be withdrawing that host from the project. I expect I will soon have similar difficult decisions to make regarding my other systems, as even their successful results are earning only small fractions (often as low as ten percent) of the credit claimed.
                                                          ____________

                                                          Profile jamin
                                                          Avatar
                                                          Send message
                                                          Joined: Nov 20 06
                                                          Posts: 324
                                                          Credit: 100,110
                                                          RAC: 0
                                                          Message 5424 - Posted 9 Jan 2007 14:04:11 UTC

                                                            Last modified: 9 Jan 2007 14:07:42 UTC

                                                            @Odysseus - or you could exchange all your 4 non-AMD boxes for 4 AMD powered ones :D

                                                            ...ohh - forgot - and install Windoze (don\'t even think of any Linux on them ;))
                                                            It will see you speeding up :D
                                                            ____________
                                                            Bang! Zoom! Right in the kisser!

                                                            Odysseus
                                                            Avatar
                                                            Send message
                                                            Joined: Feb 27 06
                                                            Posts: 212
                                                            Credit: 221,397
                                                            RAC: 0
                                                            Message 5427 - Posted 9 Jan 2007 18:06:13 UTC - in response to Message 5424.

                                                              Last modified: 9 Jan 2007 18:10:04 UTC

                                                              @Odysseus - or you could exchange all your 4 non-AMD boxes for 4 AMD powered ones :D

                                                              My AMD system (running WinXP Pro) isn’t doing any better than the Intel or the Macs; its most recent result got a “Maximum CPU time exceeded” error after nearly 180 CPU-hours. I don’t recall its receiving any credit since late in the fall.
                                                              ____________

                                                              Profile MoT
                                                              Send message
                                                              Joined: Nov 3 05
                                                              Posts: 9
                                                              Credit: 216,853
                                                              RAC: 0
                                                              Message 5428 - Posted 9 Jan 2007 21:23:47 UTC - in response to Message 5400.

                                                                This wu is not on my computer. List seems fake.

                                                                Only a week this one been pending since 24 Nov 2006
                                                                http://szdg.lpds.sztaki.hu/szdg/result.php?resultid=172906[/quote]

                                                                ____________

                                                                Profile jamin
                                                                Avatar
                                                                Send message
                                                                Joined: Nov 20 06
                                                                Posts: 324
                                                                Credit: 100,110
                                                                RAC: 0
                                                                Message 5432 - Posted 10 Jan 2007 16:35:38 UTC - in response to Message 5427.

                                                                  My AMD system (running WinXP Pro) isn’t doing any better than the Intel or the Macs; its most recent result got a “Maximum CPU time exceeded” error after nearly 180 CPU-hours. I don’t recall its receiving any credit since late in the fall.


                                                                  That\'s strange - going through my results I always see some AMD 2-3 times faster than my PentD or Core Duo :)

                                                                  ____________
                                                                  Bang! Zoom! Right in the kisser!

                                                                  Gaurav
                                                                  Send message
                                                                  Joined: Sep 4 06
                                                                  Posts: 22
                                                                  Credit: 52,275
                                                                  RAC: 0
                                                                  Message 5537 - Posted 24 Jan 2007 1:55:19 UTC

                                                                    I have a wu which after 40hrs was 60% completed, after 50hrs it went down to 20% then at 60hrs it was 40% completed, at 62hrs it reverted to 10% completed; now at 89hrs it is 50% completed.Is this an error and should I abort this wu?
                                                                    There was another wu which was I think 240% completed and reverted to 10% but it is completed now WU. My computer is Pentium D 2.8ghz running windows XP pro SP2 512mb RAM. Boinc version 5.4.11. It is running application W2.04.

                                                                    ____________

                                                                    Profile XJR-Maniac
                                                                    Avatar
                                                                    Send message
                                                                    Joined: Feb 9 06
                                                                    Posts: 27
                                                                    Credit: 225,973
                                                                    RAC: 0
                                                                    Message 5569 - Posted 26 Jan 2007 19:40:39 UTC

                                                                      Have an invalid W2.04 WU here with Maximum CPU time exceeded.

                                                                      BTW do I still have to report invalid results or is there a script running that grants some credits for invalid results?

                                                                      I\'m not crunching for credits only, but 499795 seconds of CPU time for nothing is a little bit frustrating, isn\'t it?
                                                                      ____________

                                                                      Philip Martin Kryder
                                                                      Send message
                                                                      Joined: Apr 3 06
                                                                      Posts: 73
                                                                      Credit: 10,801
                                                                      RAC: 0
                                                                      Message 5572 - Posted 27 Jan 2007 7:42:06 UTC

                                                                        Last modified: 27 Jan 2007 7:44:38 UTC

                                                                        Here is some very good NEWS
                                                                        regarding work unit 12511
                                                                        http://szdg.lpds.sztaki.hu/szdg/workunit.php?wuid=12511

                                                                        I finished it with computer 16967 on December 10 using V2.02
                                                                        See result 143610.
                                                                        Note multiple restarts.


                                                                        While computer 4937 finished on Jan 26 using V2.04
                                                                        with multiple restarts
                                                                        see result 435459

                                                                        The third computer that was granted credit also used V2.02 and had multiple restarts.

                                                                        Note also that V2.04 seems much faster (based on the relative speeds of the two machines).


                                                                        ____________

                                                                        Stick
                                                                        Send message
                                                                        Joined: Jun 12 06
                                                                        Posts: 193
                                                                        Credit: 66,244
                                                                        RAC: 0
                                                                        Message 5578 - Posted 28 Jan 2007 18:09:38 UTC - in response to Message 5572.

                                                                          Last modified: 28 Jan 2007 18:10:23 UTC

                                                                          I have seen that too - with 50 liners. But, not with 5 liners (or 2 liners). I am guessing here, but maybe the checkpointing bug is an overflow issue. That is, maybe the calculations for 1, 2 or 5 line WU\'s tend to generate data arrays that exceed the capacity of the checkpointing algorithm (and, 50 liners don\'t).

                                                                          Here is some very good NEWS
                                                                          regarding work unit 12511
                                                                          http://szdg.lpds.sztaki.hu/szdg/workunit.php?wuid=12511

                                                                          I finished it with computer 16967 on December 10 using V2.02
                                                                          See result 143610.
                                                                          Note multiple restarts.


                                                                          While computer 4937 finished on Jan 26 using V2.04
                                                                          with multiple restarts
                                                                          see result 435459

                                                                          The third computer that was granted credit also used V2.02 and had multiple restarts.

                                                                          Note also that V2.04 seems much faster (based on the relative speeds of the two machines).



                                                                          ____________

                                                                          Philip Martin Kryder
                                                                          Send message
                                                                          Joined: Apr 3 06
                                                                          Posts: 73
                                                                          Credit: 10,801
                                                                          RAC: 0
                                                                          Message 5580 - Posted 28 Jan 2007 21:33:46 UTC - in response to Message 5578.

                                                                            I have seen that too - with 50 liners. But, not with 5 liners (or 2 liners). I am guessing here, but maybe the checkpointing bug is an overflow issue. That is, maybe the calculations for 1, 2 or 5 line WU\'s tend to generate data arrays that exceed the capacity of the checkpointing algorithm (and, 50 liners don\'t).

                                                                            Here is some very good NEWS
                                                                            regarding work unit 12511
                                                                            http://szdg.lpds.sztaki.hu/szdg/workunit.php?wuid=12511

                                                                            I finished it with computer 16967 on December 10 using V2.02
                                                                            See result 143610.
                                                                            Note multiple restarts.


                                                                            While computer 4937 finished on Jan 26 using V2.04
                                                                            with multiple restarts
                                                                            see result 435459

                                                                            The third computer that was granted credit also used V2.02 and had multiple restarts.

                                                                            Note also that V2.04 seems much faster (based on the relative speeds of the two machines).



                                                                            this is just speculation - but another interpretation might be that a missing line on a 50 line wu doesn\'t move the result outside the error bounds...

                                                                            ____________

                                                                            Gaurav
                                                                            Send message
                                                                            Joined: Sep 4 06
                                                                            Posts: 22
                                                                            Credit: 52,275
                                                                            RAC: 0
                                                                            Message 5696 - Posted 12 Feb 2007 2:26:13 UTC

                                                                              I just aborted this wu which was 80% completed in 40hrs but suddenly it went down to 20% completion. I am running v2.4 on my computer. I have recently got these type of errors on at lest 4-5 wus.
                                                                              ____________

                                                                              Alex Plantema
                                                                              Send message
                                                                              Joined: Jul 16 05
                                                                              Posts: 10
                                                                              Credit: 5,221
                                                                              RAC: 0
                                                                              Message 5841 - Posted 5 Mar 2007 18:32:26 UTC

                                                                                Today I received credit for this 1-line work unit, which took 107 hours to complete. I was the only one who returned a non-empty result, but everyone else received the same credit. Isn\'t that strange?
                                                                                ____________

                                                                                Ironheart
                                                                                Send message
                                                                                Joined: Jan 16 07
                                                                                Posts: 20
                                                                                Credit: 4,366
                                                                                RAC: 0
                                                                                Message 5843 - Posted 5 Mar 2007 21:29:32 UTC

                                                                                  Last modified: 5 Mar 2007 21:43:03 UTC

                                                                                  okay, okay ... I\'m a \"beginner\" here ... but what is with the task continuing to use CPU time and the time to completion just growing larger and larger while the progress is stalled at 20%?
                                                                                  Not complaining just curious if this is a symptom of some general error? (perhaps the one being discussed most recently)
                                                                                  When I look at the results I\'m seeing that they suggest that \"1a9d6387-6dd8-4e3d-a31c-ae75a2daaf6d_61caa300-d94b-4141-871a-89a88a5d62d9_8031_1\"
                                                                                  Client detached ... was this when I lost power the other night?
                                                                                  I don\'t see any other WU\'s assigned to my account so am curious if this is part of the \'Bug\'
                                                                                  ____________

                                                                                  Stick
                                                                                  Send message
                                                                                  Joined: Jun 12 06
                                                                                  Posts: 193
                                                                                  Credit: 66,244
                                                                                  RAC: 0
                                                                                  Message 5844 - Posted 5 Mar 2007 21:46:31 UTC - in response to Message 5843.

                                                                                    Last modified: 5 Mar 2007 21:52:16 UTC

                                                                                    okay, okay ... I\'m a \"beginner\" here ... but what is with the task continuing to use CPU time and the time to completion just growing larger and larger while the progress is stalled at 20%?
                                                                                    Not complaining just curious if this is a symptom of some general error? (perhaps the one being discussed most recently)
                                                                                    I don\'t see any other WU\'s assigned to my account so am curious if this is part of the \'Bug\'


                                                                                    WU\'s are composed of lines (of matrices) to be calculated (1, 2, 5, 10 or 50). Progress moves ahead in discrete units corresponding to the number of lines in the WU. Time to completion doesn\'t decrease intra-line. My guess is you have a 5 liner that is taking a long time to process line 2. As long as you don\'t restart your computer while it is processing, you\'ll be OK. If you do have to restart, you may encounter the checkpointing bug - causing the program to jump ahead to the next line (and probably invalidating the result).

                                                                                    ____________

                                                                                    Ironheart
                                                                                    Send message
                                                                                    Joined: Jan 16 07
                                                                                    Posts: 20
                                                                                    Credit: 4,366
                                                                                    RAC: 0
                                                                                    Message 5845 - Posted 5 Mar 2007 23:12:43 UTC - in response to Message 5844.

                                                                                      okay, okay ... I\'m a \"beginner\" here ... but what is with the task continuing to use CPU time and the time to completion just growing larger and larger while the progress is stalled at 20%?
                                                                                      Not complaining just curious if this is a symptom of some general error? (perhaps the one being discussed most recently)
                                                                                      I don\'t see any other WU\'s assigned to my account so am curious if this is part of the \'Bug\'


                                                                                      WU\'s are composed of lines (of matrices) to be calculated (1, 2, 5, 10 or 50). Progress moves ahead in discrete units corresponding to the number of lines in the WU. Time to completion doesn\'t decrease intra-line. My guess is you have a 5 liner that is taking a long time to process line 2. As long as you don\'t restart your computer while it is processing, you\'ll be OK. If you do have to restart, you may encounter the checkpointing bug - causing the program to jump ahead to the next line (and probably invalidating the result).


                                                                                      Thank you very much ... with my dependence on 4 little panels for solar electric I often find myself restarted without a proactive decision ... my suspicions were correct, though my knowledge of the specific code was/is limited.
                                                                                      ____________

                                                                                      Gaurav
                                                                                      Send message
                                                                                      Joined: Sep 4 06
                                                                                      Posts: 22
                                                                                      Credit: 52,275
                                                                                      RAC: 0
                                                                                      Message 5913 - Posted 15 Mar 2007 4:04:39 UTC

                                                                                        why 69 granted in this wu - http://szdg.lpds.sztaki.hu/szdg/workunit.php?wuid=116127 when all computers that crunched the wu sucessfully claimed 200+ ?

                                                                                        ____________

                                                                                        Profile [BAT] Deridexki
                                                                                        Send message
                                                                                        Joined: Sep 13 05
                                                                                        Posts: 1
                                                                                        Credit: 233,948
                                                                                        RAC: 0
                                                                                        Message 5995 - Posted 27 Mar 2007 19:53:35 UTC

                                                                                          i\'ve attached (again) 2 dual opteron systems (24/7) to the sztaki project. I was hoping that credit granting / pending credits etc. was kind of being fixed with new core client apps being released.
                                                                                          unfortunatly i notice that it still is a complete mess...

                                                                                          i\'ve noticed workunits with pending credits of 400+ get 0.00 granted because of \"to many results\" ... (ex. http://szdg.lpds.sztaki.hu/szdg/workunit.php?wuid=56802 ) (not that difficult with workunits created dec 2006 still being handed out with a \"pending\" status)

                                                                                          why are we being tortured with this :( it\'s a complete waste of cpu cycles.
                                                                                          ____________

                                                                                          Post to thread

                                                                                          Message boards : SZTAKI Desktop Grid : New application W2.03/W2.04 - Feedback


                                                                                          Home | My Account | Message Boards


                                                                                          Copyright © 2017 SZTAKI Desktop Grid