It appears that BOINC has an issue with regards to saving completed work progress on a specific task after having rebooted a computer for updates/etc.. As we well know regardless of what operating system one uses, a system reboot is a necessary for keeping the said system up to date. Therefore I propose a system by which a user/contributor can initiate a "save task progress to the HDD" prior to any system reboot. I have seen this problem where no matter the platform or project when a reboot is done there are times where a task's progress has been reset to 0% or fails altogether. This is unacceptable, and I think having this feature to write task progress to disk before a system reboot occurs would help stem the tide of lost research and wasted participant time. As I would have it done a participant would simply have the option available in BOINC to do so before doing a system reboot. Would be nice guys! Surprised this feature has not been implemented before...

All SETI@Home tasks write periodic 'checkpoints' to disk, for precisely this purpose. After a reboot, the SETI application can read back the 'progress so far', and continue from close to where it was interrupted.

If you haven't changed any optional settings, these checkpoints are written once every 60 seconds or thereabouts: you should never need to lose more than 60 seconds of work per task when you reboot.

Other BOINC projects may behave differently. Not every project application can write a checkpoint file 'on demand', to meet that fixed timetable. Sometimes, they have to wait for a coherent point in the processing when a full set of files can be written - this may happen less often than once a minute. Other projects may have found it impossible to implement a checkpointing function, and their tasks would behave in the manner you describe (reverting to zero after every restart).

But in general, the BOINC checkpointing process works well at this project. If your experience is different, please start a 'problem' thread in the area for the operating system you're using.