http://defect.opensolaris.org/bz/show_bug.cgi?id=5444
--- Comment #4 from Aubrey.Li <aubrey.li at intel.com> 2008-12-17 06:25:28 ---
Hi Bill,
(In reply to comment #3)
> Changing all of the defines (CSTATE_INTERVAL and watermarks) to variables
> would allow tuning of this for different policies.
Better idle, :)
>
> It might be better to remove the if (mcpu->cstate_interval++ == 0) check?
> The case is only true the first time through or after a long time.
> Instead cstate_history and cstate_ep can be initialized to values to allow
> processors to only go into C1 until they have gone through one complete
> interval.
>
> + /*
> + * record the time stamp of the first time cstate entry point
> + */
> + if (mcpu->cstate_interval++ == 0) {
> + mcpu->cstate_ep = start;
This is true.
Do we decide to use this mechanism to be c-state policy?
If so, please feel free to improve it.
I more concern the behavior on EP platform(mine is a desktop platform).
How does the current interval and watermarks work on your side?
Thanks,
-Aubrey
--
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.
