In BOINC manager when a W/U is running the lapsed time naturally increases. I don't understand why the time remaining increases instead of reducing. Anybody have an answer

The "completion time" estimate supplied by the project is just a guess, and the actual time ON YOUR COMPUTER will differ from this.
BOINC is trying to work out the real figure, and will get there eventually. The more WUs that it completes, the more accurate it will be near the start.
The time WILL start to go down, but how long this will take depends on the WU. The longest climate models may take weeks to settle down.

Thanks Les, but i feel sure that the projected time should be a lot better than a guess !
SETI knows the exact size of the W/U and also knows the size of my CPU and RAM. After running benchmark tests SETI, has a fairly accurate assessment of the the projected time that my computer will need to complete. This has always been the case in the past. I think that i am losing out on this W/U as it shouldn't take nearly twice as long. it's a case of GIGO in this instance.

Seti has tasks of variable length, so if your computer has done a lot of shorties (7 day deadline), it's used to that. Then you get a long one in (deadline 4 weeks) and it'll be reassessing where it'll end.

And that's for the Seti Enhanced tasks only. The Astropulse tasks take way way longer.

As for the estimated time, it's just that. Perhaps Seti should do it better than what you're seeing, but then what about the next person with a whole different computer? Or do you think Seti has a database full of all right guesses for all the different CPU/OS combinations out there for all the different tasks that they have? Of course they don't, so an estimate is all they can send along. :-)

---

Thanks Les, but i feel sure that the projected time should be a lot better than a guess !

Even though it's not the case for SETI, many projects need to run a task to know exactly how long it will run...

---

Thanks Les, but i feel sure that the projected time should be a lot better than a guess !

Even though it's not the case for SETI, many projects need to run a task to know exactly how long it will run...

Seti provides a fairly good estimate for the most WU, it does have the additional complication on system that use the GPU since WU of the same AR will run much faster on the GPU then on the CPU.

Boinc does have a DCF (Duration Correction Factor) for each project and it tries to adjust the project estimate to provide a more accurate completion time for your computer. This is a single value that is used across all the applications for a given project. Seti has only two applications so most of the time Boinc copes fairly well. Other projects like WCG have a half dozen different applications, the estimates can vary greatly in accuracy. Other project can only give a best guess for a batch and every now and then you get a WU that is way off from the rest which can throws the DCF out of whack.

The devs are looking at the possibility of keeping a separate DCF for each application in each project as well as for the GPU. This will add a lot programming complications when it come time to clear the data when an app becomes obsolete or a new app is introduced.

---

Boinc V 7.4.36
Win7 i5 3.33G 4GB NVidia 470