I have noticed, that tasks going into "high priority" mode is dependent on the work buffer.Set low (0.1 days)--no task in high priority, setting it to 10 days, a number of tasks are going into "high priority" mode.(No additional tasks have been downloaded.

What does the work buffer setting have to do with that???

The only thing relevant are the existing tasks, not what i am requesting.

I am using "BoincTasks"(boinc 7.6.6) which actually shows tasks in "high priority mode".

Any explanation?

When a unit gets close to expiration, it goes into High Priority mode. This depends on how fast you turn over units. You can always set your buffer low, but even at maximum, you get 100 tasks for the CPU, regardless of the number of cores, and 100 per individual GPU. I would guess that your high priority units are being done with your CPU, which usually takes much longer than a GPU does. With a 10 day buffer, you still have a maximum of 100 tasks for the CPU, but it might not be able to finish them fast enough to beat the deadline. AP units usually take even longer on a CPU, so that may be a part of it also.

My CPU will completely empty itself between 2 or 3 days, with a fresh cache of 100 units. The GPU's will empty themselves in just a few hours. Often I go dry on the GPU's Tuesday during the outage.

Steve

---

Warning, addicted to SETI crunching!
Crunching as a member of GPU Users Group.
GPUUG Website

Yes. "Store at least xx days of work" has the unspoken connotation 'because I may not be able to get online again until then' - think a laptop being taken on a trip with uncertain internet access.

Any task already on the machine with a deadline closer than xx days is at risk of missing deadline if it falls due during that potential offline period.

So, BOINC thinks it's safer to get the task(s) with close deadlines cleared off the system as quickly as possible while it can - that's all 'high priority' means, do the jobs with the closest deadlines first.