Hi people !!

I'm having some trouble getting work on my home computer (AMD Athlon 2500+, mb Asus A7V8X, RAM 512 MB, HD Maxtor 80 GB, GeForce FX5200, Win XP Pro, dial up connection). I'm running BOINC 4.45 and Einstein, Predictor, LHC and SETI, and preferences are set to run always.

Because of dial-up, I have set the connection interval to four days (345600 sec), but application servers send just between one to three WU's, so my computer always has less than a day before running out of work; sometimes it won't connect even having less than an hour of work left.

Some of the messages I usually get are:

13/09/2005 08:31:52 a.m.||Suspending work fetch because computer is overcommitted.

13/09/2005 04:38:18 p.m.||Using earliest-deadline-first scheduling because computer is overcommitted.

14/09/2005 07:55:00 a.m.||May run out of work in 4.00 days; requesting more


LHC gives me more detailed explanation:

17/08/2005 08:01:02 a.m.|LHC@home|Message from server: (won't finish in time) Computer on 45.1% of time, BOINC on 45.3% of that, this project gets 36.4% of that

14/09/2005 07:55:06 a.m.|LHC@home|Message from server: (won't finish in time) Computer on -15.6% of time, BOINC on -16.7% of that, this project gets 28.6% of that


Note that I'm now getting negative figures !!!

Thing is, my computer is 24x7 on (I only reset it twice a week and give it a "free night" maybe once a month), but it only get enough work for 12 to 16 hours (at best!!), so runs out of it very often.

Do any of you guys know what's happening and what can I do to fix it?

I think I've maybe seen something similar in another thread some time ago, but I can't find it, so sorry if I'm asking for something already answered (and for this long description!!).

Thanks a lot !!
Best wishes,
Dario.

---

Wrong forum for this type of problem you need to check in at the projects forum...

---

BOINC Wiki

Wrong forum for this type of problem you need to check in at the projects forum...

Thank you for your answer, Jim.

I thought that, as this problem involves all projects, should be something related to BOINC itself or my home computer (it doesn't happen on my office PC).

Don't you think there may be one central reason for this behavior, instead as many causes as projects I'm in?

Thanks again.
Dario.

---

Boinc 4.45 doesn't estimate the correct times of the work units yet. It doesn't learn, as 4.72 and above do. So if you get 4 days worth of Seti units, in essence you do get 4 days worth of Seti, if you add up all the begin estimates of the Seti units together. That they run faster on your PC is the problem that we're trying to fix in 4.72 and above.

Having 4 projects with 4 days of work does bring your computer to overcommittence. With the new work scheduler in 4.45 and above, once your computer is overcommitted, it won't fetch new work. Once you have too many work units for all projects to run on the old time estimates, it will take the project with the nearest to deadline units and crunch those first.

That your projects are shown in negative % form may have to do with one project constantly going into EDF mode, where the rest is no longer able to run their units, until they of course near their deadline (last 24 hours before deadline). At which moment they will start to run.

As for the fix.. well, wait until 5.2.0 is mainstream. That means you will have to wait for us testers to iron out all the 5.1.x bugs first. Or you download and install 4.72 from here. Uninstall 4.45 first though.

---

OK, I'll install and try 4.72 at home while waiting for 5.20.

Thanks a lot for your explanation, Ageless.

---

Do know that it will take a while for 4.72 to learn. If at all it ever stops learning, but okay. The files coming in will still have the same estimates as 4.45 did, but once you have crunched and returned them, the next bunch should have the new estimate. At any next crunch the estimates should go down by some 10%.

---

I've been on 4.72 for a while now (July 28th I think) and am having the same problems keeping work queues full.

I'm currently connected to three projects, but two are not currently issuing work and are Suspended and set to No New Work. They all have equal resource share, and my queue size is set to 4 days.

I am getting messages that the computer is overcommitted with only 6.75 hours of work. Tonight it crunched through its last WU and got only another 3 WUs with a Time to Completion of 2:17 each. Hardly 4 days of work, or even 50% or less of the Time between Connection setting.

What's up?

---