LHC@home

With 16 cores I am currently using only 4 for ATLAS.
But what is the best use for a single task?
4 or 6 or 8?

How should the CPU% be showing while the task is working?
I see a high percentage on start up up to .70% of 4 cores and then if I come back in an hour or so it drops to .03% of 4 cores and is clicking off about .10% completion every 2 seconds (BoincTasks monitoring).

What is it doing? Simulated runs first and then data analysis afterwards?
I am trying to understand the process so I can better tell when a task is stuck (a .001% per 2 seconds completion time usually marks a stall) or when it is running normal. What does it do after the first so called 15 seconds of run time vs elapsed time of 4 hours plus to complete the task?

And now after 7 hours run time on 4 cores it bogs down.
Now only moving forward at .004% per 2 seconds on .02% cpu power.

Stderr file output:
2021-05-01 21:12:02 (19856): Setting CPU throttle for VM. (100%)
2021-05-01 21:12:02 (19856): Setting checkpoint interval to 900 seconds. (Higher value of (Preference: 180 seconds) or (Vbox_job.xml: 900 seconds))
2021-05-01 23:18:28 (19856): Status Report: Elapsed Time: '12802.115137'
2021-05-01 23:18:28 (19856): Status Report: CPU Time: '19.859375'
2021-05-02 01:12:39 (19856): Status Report: Elapsed Time: '18802.115137'
2021-05-02 01:12:40 (19856): Status Report: CPU Time: '23.484375'

This stinks.

what is the best use for a single task?

One.

yeah i might have to try that being i blew up a bunch of tasks already with 4 and 8.
I used to be able to run ATLAS with 4 no problem, its 50/50 now with 4.