Yeah mine does this too, I even binned off the Windows agent program and tried it through Boinc and it still starts over back at zero after i stopped and restarted boinc, I am throughly annoyed by this cos it makes me wonder whats the point of doing this project if all the work counts for nothing every time we shut down our PC's. Some of us dont have machines that can stay on 24/7 so surely any program worth its salt should be able to save its current progress to the hard disk, surely thats not too much to ask.

I've even let the project get as far as 25% the way along one of its current jobs and guess what i restarted the PC and back to zero it went. VERY FRUSTRATING !!!!

FAAH is short for FightAIDS@Home... one of our currently running research projects.

The research programs do checkpoint their intermediate results at certain times, writing this checkpoint information to the disk. If the machine is turned off or shut down, the agent would resume from the last checkpoint. Due to the nature of the research code, sometimes the checkpoint opportunities can be more than an hour apart.

If you have other applications using a lot of cpu time, or your machine is not particularly fast, the checkpoints are usually further apart in time. So, if you are unfortunate enough to stop it just before the next checkpoint, the work since the last one or since the beginning of the run can be lost.

One way to avoid this is to make use of hibernation, standby, or suspend modes instead of shutting down your PC. It depends on what kind of PC you have, whether these options are supported. These options would be listed on your Windows Start-> Shutdown menu, if available and enabled. They effectively would make the agent never lose any of the work because the system can resume running from exactly where it was, leaving all of your other applications open as well when you resume processing. A system crash, or ungracefully cutting power off to the machine would however lose the work since the last checkpoint.

The code hasn't changed.

This has been answered so many times today, I have to ask: does anyone bother to read the other posts before asking their question anymore? It's Netiquette 101.

Anyhow, we don't mind answering questions no matter how often they are repeated.

What has changed is the work units themselves. Clearly, the current batch has fewer checkpoints than the average. Far fewer. None, sometimes. WCG don't really get any extra information about work unit batches beforehand, or they might have been able to lessen the effect of all these unbroken work units.

Just a suggestion-

Let the user to choose which type(s) of WUs for crunching in the User Profile page so that those people have to shut down their PCs in a day can take those WUs with more check points whilst those with 24/7 PCs can crunch those with no or few check points.

At present, it just like working in a black box without knowing which types of WUs are receiving/under crunching. Also, there seems no correlation between the Task Run Time and Task Execution Progress and there's no way to tell how long the PC will hit the next check point. This is one of the un-userfriendly program that I've seen.