Problem on AMD64

Alexandre "Sunny" Kovalenko gaijin.k at gmail.com
Thu Jan 1 17:09:56 UTC 2009


On Sun, 2008-12-28 at 13:20 +1100, Ian Smith wrote:
> On Sat, 27 Dec 2008, Alexandre "Sunny" Kovalenko wrote:
>  > On Sat, 2008-12-27 at 17:13 +1100, Ian Smith wrote:
>  > > On Fri, 26 Dec 2008, Garrett Cooper wrote:
>  > >  > On Thu, Dec 25, 2008 at 1:14 AM, David van Kuijk <dynasore at bigfoot.com> wrote:
>  > >  [..]
>  > >  > >> Look into the following sysctls:
>  > >  > >>
>  > >  > >> hw.acpi.cpu.cx_lowest
>  > >  > >> hw.acpi.cpu.cx_highest
>  > >  > >>
>  > >  > >
>  > >  > > I tried to find out what I can do with those sysctls.
>  > >  > > hw.acpi.cpu.cx_highest is not available on my system.
>  > > 
>  > > I don't think it exists; C1 state is always available AFAIK.
>  > > 
>  > >  > > hw.acpi.cpu.cx_lowest is available and can be set. Are you suggesting I
>  > >  > > should set it to C2 or C3???
>  > > 
>  > > Give it a try.  Setting it to C3 won't hurt, whether it's used or not.  
> 
>  > Actually it just might (hurt that is). See below for the discussion.
>  > 
>  > http://www.freebsd.org/cgi/getmsg.cgi?fetch=0+2482
>  > +/usr/local/www/db/text/2008/freebsd-acpi/20081109.freebsd-acpi
> 
> Thanks.  Yes that was indeed an interesting discussion.  FWIW, I find 
> that interface awkward, without threading.  For me it's easier from:
> http://lists.freebsd.org/pipermail/freebsd-acpi/2008-November/005182.html
> 
> Could I then safely say, setting it to C2 won't hurt?
It is hard to generalize. I am using:

RabbitsDen# sysctl dev.cpu.1.cx_lowest=C3
dev.cpu.1.cx_lowest: C2 -> C3
RabbitsDen# sysctl dev.cpu | grep cx
dev.cpu.0.cx_supported: C1/1 C2/1 C3/57
dev.cpu.0.cx_lowest: C2
dev.cpu.0.cx_usage: 0.00% 100.00% 0.00%
dev.cpu.1.cx_supported: C1/1 C2/1 C3/57
dev.cpu.1.cx_lowest: C3
dev.cpu.1.cx_usage: 0.00% 30.43% 69.56%

without any obvious ill effects. Whether having C3 on one of the cores
is makes any difference in power consumption or heat dissipation, I do
not know.

I also do not know whether there are machines out there which would
exhibit odd behavior with C2 set.

As usual, YMMV.
-- 
Alexandre "Sunny" Kovalenko (Олександр Коваленко)



More information about the freebsd-acpi mailing list