I only use BOINC to run World Community Grid (Registered Member Since April 2006). I have never run any other BOINC project (Einstein, SETI, etc.). This situation is related to https://www.worldcommunitygrid.org/ms/viewMyProjects.do which looks like this: ~~~~~~~
| Available Projects | | All | None | | | | | Smash Childhood Cancer | | | | | OpenZika | | | | | Help Stop TB | | | | | FightAIDS@Home - Phase 2 | | | | | Outsmart Ebola Together • Please review the system requirements before opting to participate in this project | | | | | Mapping Cancer Markers | | | | | FightAIDS@Home | | | | Please note: Some projects may occasionally have periods where no work is available. If you choose to receive work for only one project, the project may, at times, not have work units available for download. For this reason, we recommend that you select more than one project. | | | | | | | | If my computer can process work on my graphics card, then please send me work to run on my graphics card for the projects that I have selected above. | | | | | | If there is no work available for the project(s) I have selected above, please send me work from another project. | | | | | | Please opt me in to new projects as they become available. | Changes will not take effect until the next time that your grid agent communicates with the servers (i.e., when the current work unit has finished processing~~~~~~~ WCG offers 7 projects, of which 3 are currently selected. The last checkbox refers to the possibility of sending work from one or more of the 4 WCG projects that are not currently selected when no work is available from any of the three projects that are selected, or possibly when no work is available from at least one of the three projects. The assumption that work would be unavailable for one day per week was intended only to make it easy to illustrate the problem with the BOINC fetching algorithm as applied to WCG. It does not relate to the actual pattern of work outages at World Community Grid. The 7-day buffer does not operate only as insurance against WCG work outages. It also insures against (local network and/or ISP) Internet outages. WCG itself (the whole thing) recently crashed for nearly a week. BOINC provides an "Abort Task" button without limitation, and there should be no philosophical problem with using this function, either manually or automatically. The algorithm I outlined as being the proper way to handle the situation I first described involves some coordination between BOINC and WCG. I interpret your statement "BOINC doesn't recognize tasks though, it works with seconds of work to download" to mean that BOINC has no knowledge of the WCG project preference settings as indicated at https://www.worldcommunitygrid.org/ms/viewMyProjects.do . Is that correct? From: Jord van der Elst <[email protected]> To: William <[email protected]> Cc: BOINC Developers Mailing List <[email protected]> Sent: Friday, July 14, 2017 9:53 AM Subject: Re: [boinc_dev] Request re: Fetch New Work On Wed, Jul 12, 2017 at 10:43 PM, William <[email protected]> wrote: > Proper procedure would be to recognize that work is now available from the > selected project(s), abort all those "Ready To Start" tasks from the > unselected project(s), then calculate how many new tasks are now needed from > the selected project(s) and fetch work from the selected project(s) > accordingly. If everyone would be requesting a full 7 days worth of work to fill in the time that the original projects don't have work for one day, then abort that whole cache of work, none of the work will ever be done. It's also not how this function works. You have the option to either have your computer be partially or completely idle during the time that the original project is in maintenance and cannot send you work, or for your computer to be busy doing at least some good work for the other project(s). You chose the last. If anything should be done, it would be to download enough tasks to fill the CPUs/GPUs and nothing more, so as not to waste when the original project comes back. And then download one task every time one task is finished. BOINC doesn't recognize tasks though, it works with seconds of work to download. With a 7 day cache that's a lot of seconds, even for the backup project. Also, the option "If there is no work available for the project(s) I have selected above, please send me work from another project" isn't so much one of the unselected projects in WCG (although it may be working like that in this case), but another project you may have added to BOINC (Einstein, Seti, CPDN, GPUGrid, etc.) > Requesting that this correct procedure for handling the World Community > Grid's checkbox related to work outages be included within an upcoming > release of the BOINC software. Yes, that's not going to happen, because it's wasteful and against the idea behind BOINC (try to finish all work before deadline). A simpler solution for you would be to add an 8 day cache, so that day that BOINC downloads work from another project only happens once every 6 weeks, and even then maybe not. Or if this is a BOINC 7.6, to set the first value, which is the low water mark, to 0.5 days and the additional work to 8 days, BOINC will download enough work to top off the cache every half day. ("Connect to network about every N days; Cache N extra days of work") It will then on the day of the outage download enough work from the backup project to top off the cache. Yes, that work still has to be done, but at least it isn't a full cache. -- Jord van der Elst. _______________________________________________ boinc_dev mailing list [email protected] https://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev To unsubscribe, visit the above URL and (near bottom of page) enter your email address.
