http://defect.opensolaris.org/bz/show_bug.cgi?id=4607
--- Comment #4 from Bill Holler <bill.holler at sun.com> 2008-11-07 11:53:08
---
(In reply to comment #3)
> Yes. I am suggesting using a different capacity for Idle Power PG.
Idle Power PG could be dynamically calculated based on the idle states of
CPUs in the Idle Power PG. Deeper idle states have less capacity while
shallower idle states have higher capacity.
Another way to calculation Idle capacity is:
PG Idle capacity = ((num CPUs in C1) << 16) + ((num CPUs in C2) << 8)
+ (num CPUs in C3)) /* 8 bits in NCPU */
CPUs can add/subtract their capacity when going idle or waking up from idle.
Capacity == 0 means no CPUs in the PG are idle.
It is not known if dynamically calculating Idle capacity will have a higher
performance penalty than the benefit.
> Active Power PG would continue to use the number of CPUs in the PG as
> capacity,
> so there will be no effect on p-state policy.
--
Configure bugmail: http://defect.opensolaris.org/bz/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.
