Work units not saving?

[S07197]

Hello everyone!

First of all, sorry if this is the wrong forum to post this. Feel free to move the topic to a more appropriate location if necessary!

So, I'm currently crunching OGR-28 on 18 or so computers, and thought I'd put my Sony Xperia Z to work on this project through yoyo.

Installing, joining & everything else went smoothly, and my phone begun crunching.
However, whenever I take the phone off the charger, the unit's progress is lost and next time I plug the phone up, it starts crunching from 0% again.
Any idea what's up with this behavior?

-S

[yoyo]

Internally the OGR application creates checkpoints roughly every 15 minutes. But this is not reported to Boinc. Therefore Boinc thinks the workunit starts from the beginning. But this is not the case, the workunit continues from the last chekpoint.

yoyo

[S07197]

Ok!

Thank you for your reply.

I'll just keep on crunching then!

-S

[S07197]

Everything seems to be working smoothly. I've already crunched plenty of work with my phone.

However, I have another small question regarding the amount of CPU time, claimed credits and granted credits...
Maybe this has been asked before and maybe the topic is in German, but here goes...
I'll just use this old topic of mine for the question.

For example, here are the results of my latest 3 WU's:

Result ID Work unit ... CPU time (sec) claimed credit granted credit
26299479 20458697 Done 3.20 0.01 10.63
26299475 20458693 Done 4.74 0.01 79.14
26299406 20458624 Done 5,207.45 11.10 57.72

First of all, how is it possible that some of the WU's only take a few seconds to complete?
Secondly, how come the work of less than 5 seconds gets much more granted credit than then work of over 5000 seconds?

Surely there is a logical reason for the very small work units and the credits, I just don't quite get it?
I don't mind the credits being funky, I'm all for the project, not the glory.. Just wondering about the mechanics behind this.

-S

[yoyo]

As already stated, the workunit itself does checkpointing and stores the state every 15 minutes. But Boinc doesn't know about it. So if Boinc is restarted it counts the CPU time from 0 and does not continue where it was before. But the workunit itself continues from the previous state.
The credit which is granted is calculated based on the executed work. So it might happen, that the workunit restarted and finished after the last restart very fast. In this case less cpu time is reported. But the credit is granted based on the whole done work.

yoyo

[S07197]

So this actually is related to the same thing I was wondering before... How convenient I posted this in the same thread.
Makes perfect sense!

Thank you again for satisfying my curious mind yoyo!
...And keep up the good work!

-S