The last in the series of project specific performance charting which was kind of a headache until it came to me while "stressing out" on the bicycle.... simply take CPU days instead of years and showing Results without a x1000 divisor.

The jacking up and down is mainly driven by the periodic release of a 14 day prediction period. With cycle 24 of 27 in progress we're now well on the home stretch after which the project goes on hiatus for analysis and improvement of a future phase II.

The increase of run time per task and dropping of points per hour is typical of the less speedy machines turning in their results a later. It's very evident particularly with the last few cycles. All the other graphs , see link in signature, show similar effects of fast versus the slower clients where points per hour will rise quickly and than drop before balancing.

enjoy and ask if something looks incomprehensible.