Why do "CPU Time" and the "To Completion" both go up? Shouldn't "To Completion" go down as "CPU Time" goes up?

---

Why do "CPU Time" and the "To Completion" both go up? Shouldn't "To Completion" go down as "CPU Time" goes up?

The initial "To Completion" time is an estimate from the project, based on your computer's benchmarks. It may be _way_ off. In order to correct it, with each result you do, the difference between the actual finished time and the original estimate is calculated, and a correction factor applied. After about 10 results, this is usually pretty accurate. Before that, BOINC Manager will _try_ to adjust the "to completion" number "on the fly" - so let me make up some numbers...

Original estimate is 10 hours. After 1 hour of crunching, % complete is only 5%, so _new_ estimate, based on actual progress, would be 20 hours. BOINC Manager increases "to completion" time by half that and says it's now 15 hours.

This is extreme, but you get the idea. It's hard for me to say "how" the number is readjusted on-the-fly because they've changed the algorithm three times in the last few versions. :-( Regardless, after 10 results, it'll be going down as you'd expect it to.

EDIT:: All of this assumes you're on a version of BOINC V4.72 or later... and a project that has results that are at least somewhat "close" to the same length... and it's not Tuesday...

---

"After about 10 results..."

That's interesting because my profile shows zero results, and I actually have thousands.

---

"After about 10 results..."

That's interesting because my profile shows zero results, and I actually have thousands.

Which project? If you're looking at your profile _here_, BOINC doesn't have results, etc., so we ALL have zero.

Also, even if you have a valid duration correction factor calculated, if the project sends a result out to you that is way longer than both their estimate and the "normal" time you take to complete one, you can see the "to completion" going up... this happens to me on Rosetta at times. Most of their WUs are pretty short, then every once in a while I get one that takes several hours instead of the normal (and estimated) two. It'll start at 2:00, then when it's only 1% done after several minutes, it'll start climbing. By the time it's at 50% or so, the estimate is fairly accurate and will start going down again.

---