The 10 day deadline is for 99.9+% sufficient, even for those that have their device on only few hours a day. Of course, this is not ideal. The problem by *not* extending deadlines is, that other server/scheduler performance issues build up. The more live entries, the slower the system becomes. Including No Reply redistributions [4 day deadline], the aim is to have most all tasks off the active system within 14 days [there's exceptions, as always for some work]. Reduction of the daily results, artificially, will be another way to allow isolating an issue, be it the BOINC system kicking back to the network or the network kicking the BOINC system into catatony.

Those [slower/low hour] devices that do have trouble, are best given the lighter jobs [HCC/HCMD2/CFSW/C4CW] by changing the My Project or Device Profile selections.

For the moment though, viewing this as a program temporary fix until these stagnations/saturations causes can be rooted out. WCG did 1.2 million tasks daily few months ago with easy, so the 925-950K now done is well within capacitance to once resolved, shorten the tasks again as was done in the past.... back to the 6 hour or shorter target average.

To those crunchers who are crunching the latest CPU-based FAAH-WUs, how is the WU going in terms of runtime?
;

I have noticed a big increase in the workunit runtime from 5 hours to the current units being 8 hours.

I have:

Processor: 2 GenuineIntel Intel(R) Pentium(R) CPU P6200 @ 2.13GHz [Family 6 Model 37 Stepping 5]

Sept 21, 1 day average 6.85 hours
Sept 21, 7 day average 6.86 hours

Crunch On.