I have always been surprised that the servers are not named after stars. Algol is a great start, Vega close behind. Rah Sagen!

But the real concern is the order in which work units are being scheduled. Ones with deadlines of next week are running, while those with a deadline of tomorrow is just sitting. I had to stop new tasks due to an OS update, and several tasks were still running after the deadline. If it is a hard deadline, I hate to waste the work if the output is rejected. If a soft deadline, why have one?

BOINC is more than just Seti@Home. If it were just Seti@Home, doing work by deadline would probably be possible, but with a lot more projects, their longer or shorter run times and their wildly fluctuating deadlines thrown into the fray, it's better to do work by first in, first out as it is done now.

That way running e.g. a Climateprediction.net model with a runtime of several months, and therefore also a deadline of several months if not a year, will not fail to be returned before its deadline, because you chose to run a project with smaller deadlines alongside it. Because when else is that CPDN model going to be run if not way too late to be able to finish it?

But similarly when you run Seti@Home's Astropulse with a month deadline and choose to run MalariaControl.net next to it, which can have 3 day deadlines. If run true by deadline only, your system will only do Malaria and maybe 24 hours prior to its deadline the Astropulse. Which would be doable if it were only one, but usually you have more than one.

---

Yes, BOINC is more than Seti@Home. But since this page and the server page say Seti@Home at the top of the page, I made the obvious assumption that I was talking to Seti and not BOINC. Oscar may be BOINC, but Paddym?

I had to pick two names, starting with G and H, for computers 7 and 8. I picked Gadfly and Humbug. Later it was Dodeca and Hedron. Eight disk drives were labeled SWhite and the Seven Dwarfs.

Of course, this is contradicted by the comment on execution order. I am, now, only running one other project, NFS@home. Can't keep the old math majors down. When it has several days worth of work and they are running "out of order" it does seem that FIFO is not working. I suppose that the project could, on a later date, submit work with an earlier date but why?

I suppose that I would monitor turnaround time, X days, and then set the deadline at 2X days. But where lieth the code? In BOINC itself, or in code written by the project.

Scheduler code is not easy. Multiple processors only make it harder. I just hate coming in a day after the deadline and having my work thrown away. If not, why have a deadline.

This is SETI but the scheduling is done by the BOINC software on your computer and at the projects. BTW, I've been at this for a long time and I'm processing several projects. The only time I've ever seen a WU miss deadline was when a project underestimated the running time by a factor of 100. So BOINC scheduling and work fetch algorithm have worked very well. BOINC is open source, if you think you can do better, you can always summit patches to the BOINC development team for consideration.

As to the naming of SETI servers, that is up to the project scientist. If you want to know how they pick names, you're welcomed to dig back a couple of years in the Technical News forum to find when it was explained.

The BOINC database is a misnomer in that it hold the SETI results. I'm not sure, but it may have been supplied as part of BOINC's server software.

Just as background information. BOINC and SETI have an incestuous relationship. BOINC was developed by the people that originally created SETI. SETI was able to continue after the initial funding dried up, in part, as the "guinea pig" project in the development of the BOINC software. Funding for SETI staff, app development, and hardware mostly comes from donations from the project volunteers.

---

Boinc V7.2.42
Win7 i5 3.33G 4GB, GTX470