Is it just me or has the estimated time remaining got worse?
Progress Elapsed Remaining (Estimated)
25.975 111.50 529:39
42.219 29.23 73:23

With the short models I have completed enough that I would have thought it should be estimating close to the real figure by now.

Not that it affects my support for the project, just curious to know if anyone else has noticed estimates getting more pessimistic?

Are you sure they are the same type of task? Your computer has one CM3 short and two RM3P-ANZ tasks in progress.

I have completed three of each of those types of task, just checked.

CM3S task just finished. Estimated time for next one gone down from 85 hours to 83 hours rather than something closer to the 63hours 29 minutes the finished one took. Seems to me that the iteration algorithm has either got worse or my memory (personal not computer) is failing.

Boinc tends to correct the estimated time remaining by about 15% for each completed model.

---

And, given the age of the BOINC server software we run here, the estimated runtime for the models is entirely calculated and maintained by your local computer - the project doesn't have any influence over runtime estimates.

The only data used to calculate the runtime estimate are the benchmark speed of your computer, the size of the task (number of floating point operations - invariant for each model type), and the 'fiddle factor' or DCF used to nudge things in the right direction when they go wrong.

Problems can arise if the task size for different models types isn't set exactly pro-rata by the project. Then, DCF will be pulled one way by one type of task, and back the other way by other types of task. With only one DCF to play with, it can't satisfy every need, and estimates might fluctuate. But we're stuck with that behaviour, and we have to just live with it.

Thanks Richard and Jim.

I will follow how the estimates change. At least with the short models they should get there reasonably quickly. And of course, not a major issue in the grand scheme of things.

Now I really don't understand what is going on, despite the last two cm3s tasks having finished in a tad over 60hours, the estimated time for an unstarted task has gone up to 89hours.

And having just moved to 7.4.22 the estimated time has dropped. However not sure whether this related to the upgrade or if things get recalculated when BOINC is stopped and restarted?