Even on new version 5.4.9 I have noticed that BOINC will download a whole set of data when first connecting to a project, assuming that no other projects would need to run until the time you have specified to connect to the network again. I have tried to change the portion percentage before downloading, but it doesn't seem to care. It also ignores what may have been sitting there from other projects, and then whines that it is "overcommitted". Well, duh!

Subsequent downloads seem to honor the portion desired for that project.

I'm not sure what this deadline change will mean, but I have reduced my own cache length from 2.0 days to 1.0 days. Hopefully things will finish sooner, now that BOINC has overcommitted some of my stuff and pushed things out.

If worse comes to worse, I can also tinker with the project screen to cut off new tasks or suspend execution of some projects so that others will finish.

I have one very antique pc which runs between 4 and 12 hours a week. If I'm lucky, it will do one result a week, but when a workunit is larger, it will take at least 16 hours or more to complete.
Will there be any kind of difference in workunit length so that slower computers or computers which are not running very often get smaller workunits?
So I mean on my "normal" computers I sometimes have workunits which complete in less than one hour. Those would take maybe 3-4 hours to complete on an old machine. That would be better.

That's a bit like trying to nail jello to a tree.

I leave my cable modem connected when I travel away and even during severe electrical storms. My philosophy is that my ISP owns the modem and if it gets damaged during an electrical storm then they will pay to replace it, not me. I do own the router behind the modem so I worry about it but not the modem.

To protect my router I have a surge supressor between the cable modem and the incoming coax cable and a second surge supressor between the modem and the router. I also have a surge supressor and noise filter on the mains outlet. I think that's the only practical solution to the problem.

Recently, I got a lot of WUs with only one week deadline. Does it mean that the allowance of 4 days for returning the results prior to re-issuing WUs is too short?

knreed,

the news on the behind the scenes extra week is good..., we'd not be contributing to the quorum points determination, but that's aside.

I've got a 'relative' fast machine in perspective to times turned in by many others. Should 1 calendar week not be enough, the sample had actually 7:08 days, probably 8 or 9 days including shut down periods, it would cause for extra copies to go out without the real need.....trust your scripts consider issues like zero returns in e.g 5/6/7 days. Anyway, if something is taking a very long time, i check on the result status detail to see what times others are turning and on basis of that determine if it makes sense to continue (with bleeding heart on the lost hours).

It's a small concern, seem to get them once every 4/5 months and generate only half the points of a regular....whilst they need to be crunched, some will be tempted to kill them.