I am using BOINC version 5.2.13 on the latest stable version of Debian. This computer\'s CPU is 832 MHz and its architecture is i686. I also have a faster Windows XP computer, running the recommended version of BOINC, and its CPU is 2 GHz.

This computer is attached to four projects: Climateprediction.net, SETI@home, Einstein@home, and LHC@home. Since LHC@home is out of work, I detached it. A few days ago, BOINC reported that it was overcommited and suspended work fetch. It also began using the earliest-deadline-first scheduling.

SETI@home had the earliest deadline. This computer finished the WU way before it. Einstein@home came next, and this computer finished that way before the deadline as well. Two days ago, the computer started the CPDN experiment and yesterday afternoon, BOINC reported it wasn\'t overcommitted anymore and fetched a workunit from SETI@home, but not from Einstein@home.

Right now, this computer is timeslicing between SETI@home and CPDN, with each of them getting a resource share of 50%. The problem is, CPDN still has 5015 hours 40 minutes in CPU time to completion, and this number is increasing, not decreasing.

The deadline is November 26, 2006. But since BOINC is timeslicing, the CPDN model will take at least 418 days to complete!

What should I do? Should I:

a) Detach Climateprediction.net and attach it to my faster computer, which is only on 25% of the time (or less)?

b) Dedicate this computer to CPDN and the faster computer to the other projects?

c) Stop running CPDN and delete my account?

For the moment, go with b. You can always change the plan later.
The deadline is not fixed. As long as the model continues to trickle to let the server know that it\'s still alive, you can go past the deadline without problems.

I agree with Les and if you don\'t want to lose SETI\' WU you can (in Boinc Manager) stop to get new work from SETI. Nevermind, I doubt you will not finish before deadline because you have 11 months to finish your CPDN WU (and I will probably take 4-6 months with your computer to finish one). The estimated completion time is probably over-estimated as well (around 2x for my laptop).

---

This is a real problem for me, and I hope the climateprediction folks lean on the BOINC client folks to fix their scheduler. There is a thread here that I started, dealing with this. It is not done yet, and IMAO some of the respondents do not understand how things actually work, but there is now some recognition that there is a problem.

http://boinc.berkeley.edu/dev/forum_thread.php?id=390

---

I agree with Les and if you don\'t want to lose SETI\' WU you can (in Boinc Manager) stop to get new work from SETI. Nevermind, I doubt you will not finish before deadline because you have 11 months to finish your CPDN WU (and I will probably take 4-6 months with your computer to finish one). The estimated completion time is probably over-estimated as well (around 2x for my laptop).

I suspended work fetch for SETI@home but for some reason not for Einstein@home. Anyways, when Einstein@home reported its result, I detached both projects. Climateprediction.net has been alone for a few days now.

The completion time for my computer is actually under-estimated. With some simple calculations, I found out that if the time is correct, CPDN would trickle every 2.8 days. The first trickle came on December 20, 6 days after work was downloaded (BOINC wasn\'t timeslicing for most of that time).

I have no idea of CPDN\'s progress right now. I changed gui_rpc_auth.cfg a few days ago but didn\'t restart the BOINC client. There\'s nothing interesting to see--frequently using the \"Show Graphics\" button would crash the model as it did the previous two times.

---