My most recent workunits were all 1852s...

1852- grandchild
1852- child
1852- child

Today's work units:

1881 - child
1804 - child repair job
1879 - grandchild

Sekerob, this is week is usually the last week for many colleges before summer break (well at least the one I went to). So all those college students stopped crunching (well at least for a while -- entire summer or just a vacation)

For exclusive HCMD2 crunchers the disparity of WU sizes often causes a device to not request/report work for long times, and for larger work queues that may lead to a full day without reporting if the device is really running without human monitoring.
Even more since I have always seen parent WUs estimated as shorter than child or grandchild ones. It might make sense for slower machines which will stop the big ones at 6 hours while some really shorter ones might be allowed to run longer exceptionally. But for fast machines (which are also those with more jobs in queue) that leads to this yoyo effect of the estimated durations and thus to these silent periods: after a series of grandchildren the estimated duration is low and more WUs are loaded, and when the first of the next series of parents completes the queue can suddenly become four times longer or more.

Sadly, anhhai will like have been on to it... all other WCG sciences followed with a dip in the afternoon... the "schools-out" is upon the crunching world.



PS, Lucky for the schoolkids here, around the 15th means not to return to the benches until September. Nearly 3 months of safer roads [not so packed with stressful mothers/some dads hauling their kids to school and back.