Gerry --

I see the FIPE driver directly modifies the throttling 
registers in the memory controller.

Although there is no issue in doing this in Intel systems 
with a 7300 or 5000 series Memory Controller Hub, I want 
bring up that OS configuring and enabling throttling 
registers in later (Nehalem-based) Intel systems should be 
avoided.  System designers assume that BIOS and/or the 
service processor have ownership of throttling behavior. 
Throttling may be used for managing system issues the OS is 
not aware of.  If the OS and the layer beneath the OS both 
attempt to use throttling for different purposes, this could 
cause a problem.

Do you see the cpu idle notification / FIPE driver solution 
as a long lived approach to memory power management?

-- jdh


Liu, Jiang wrote:
> Hi buddies,
>       I have developed a device driver for Intel 5000/7300 chipset
> which puts memory into power saving state when system is idle.
>       Could you please help to review it? Webrev is at
> http://cr.opensolaris.org/~gerry/fipe_20090408/.
>       There are two main parts in the webrev. The first one is
> enhancement to dcopy/ioat driver to support FIPE. The second one
> is FIPE driver it self.
>       Thanks!
> Liu Jiang (Gerry)
> Senior Software Engineer 
> OpenSolaris, OTC, SSG, Intel
> Tel: (8610)82171643
> iNet: 8-758-1643
> Location: Raycom 9W013
> _______________________________________________
> tesla-dev mailing list
> tesla-dev at opensolaris.org
> http://mail.opensolaris.org/mailman/listinfo/tesla-dev

-- 

---------------------
     Julia Harper, julia.harper at sun.com


Reply via email to