Question about AthlonXP on nforce 2 : C2-C3 states

Question about AthlonXP on nforce 2 : C2-C3 states

[Arthur Huillet]

Hello,

I've recently discovered that my system didn't seem to go into C2/C3 states. In fact, they're not
even recognized :

# cat /proc/acpi/processor/CPU0/info 
processor id:            0
acpi id:                 0
bus mastering control:   no
power management:        no
throttling control:      no
limit interface:         no

# cat /proc/acpi/processor/CPU0/power 
active state:            C1
max_cstate:              C8
bus master activity:     00000000
states:
   *C1:                  type[C1] promotion[--] demotion[--] latency[000] usage[00000000]

Would you happen to know why the Cx states are not seen ? Also, isn't there supposed to be bus
mastering control available ? 
My mainboard is a Asus A7N8X rev 1.xx (nforce2).
The kernels I tried with no success in whatever case are 2.4.29 and 2.6.15.2.
Unfortunately, google doesn't seem to know much about this problem...

-- 
Greetings, 
A.H.

Re: Question about AthlonXP on nforce 2 : C2-C3 states

[Pavel Troller]

> Hello,
> 
> I've recently discovered that my system didn't seem to go into C2/C3 states. In fact, they're not
> even recognized :
> 
> # cat /proc/acpi/processor/CPU0/info 
> processor id:            0
> acpi id:                 0
> bus mastering control:   no
> power management:        no
> throttling control:      no
> limit interface:         no
> 
> # cat /proc/acpi/processor/CPU0/power 
> active state:            C1
> max_cstate:              C8
> bus master activity:     00000000
> states:
>    *C1:                  type[C1] promotion[--] demotion[--] latency[000] usage[00000000]
> 
> Would you happen to know why the Cx states are not seen ? Also, isn't there supposed to be bus
> mastering control available ? 
> My mainboard is a Asus A7N8X rev 1.xx (nforce2).
> The kernels I tried with no success in whatever case are 2.4.29 and 2.6.15.2.
> Unfortunately, google doesn't seem to know much about this problem...
> 
Hi!
  Please try to do something like this:

root@porygon:~/# cat /proc/acpi/fadt |acpitbl |grep P_LVL
P_LVL2_LAT:       190
P_LVL3_LAT:       1900
If you will see more than 100 in the first line and more than 1000 in the second one, your
Cx states are blocked by BIOS. My example is from ASUS A7V8X, which may be similar.

              With regards, Pavel Troller

Re: Question about AthlonXP on nforce 2 : C2-C3 states

[Arthur Huillet]

Hi,

On Sun, 5 Feb 2006 15:37:42 +0100
Pavel Troller <patrol@sinus.cz> wrote:

> If you will see more than 100 in the first line and more than 1000 in the second one, your
> Cx states are blocked by BIOS.

I'm getting 101 and 1001 indeed : 
P_LVL2_LAT:       101
P_LVL3_LAT:       1001

Is there no way to tell my BIOS to unblock those states ? I mean, are they blocked for stability
purposes (in what case I assume this is a workaround for a bug in the chipset/CPU/whatever), or are
they just not enabled for an obscure reason ?

Thanks.
 -- 
Greetings, 
A.H.

Re: Question about AthlonXP on nforce 2 : C2-C3 states

[Pavel Troller]

> Hi,
> 
> > If you will see more than 100 in the first line and more than 1000 in the second one, your
> > Cx states are blocked by BIOS.
> 
> I'm getting 101 and 1001 indeed : 
> P_LVL2_LAT:       101
> P_LVL3_LAT:       1001
> 
> Is there no way to tell my BIOS to unblock those states ? I mean, are they blocked for stability
> purposes (in what case I assume this is a workaround for a bug in the chipset/CPU/whatever), or are
> they just not enabled for an obscure reason ?
  Wellcome to the large family of owners of boards/computers with broken ACPI
BIOSes (or hardware bugs ???)...
  I didn't find any way how to unblock them. Nobody but BIOS authors know, why
they are blocked, but it seems to be a common practice of current ACPI BIOS
writers :-(. 
  I have the same problem with ALL of my machines, including my notebook
(Clevo D610SU). None of them supports native Linux ACPI because of this.
On Athlon boards, I'm successfully using athcool
( http://members.jcom.home.ne.jp/jacobi/linux/files/athcool-0.3.11.tar.gz ),
it substantially decreases CPU temperature even with C1 due to enabling
"HALT Command detection" bit on the NB. On some systems, I've patched kernel
to use C2 and C3, which, together with athcool, seems to have even better
effect:
root@co:/home/tv/il2sturmovikfb# cat /proc/acpi/processor/CPU1/power
active state:            C2
max_cstate:              C8
bus master activity:     000007ff
states:
    C1:                  type[C1] promotion[C2] demotion[--] latency[000] usage[17803710] duration[00000000000000000000]
   *C2:                  type[C2] promotion[--] demotion[C1] latency[101] usage[116616196] duration[00000000277739332216]
You can see 101 in latency field, which is normally prohibited :-)
C3 is not used because of lack of bus mastering capability
(it's on another of my boards, MSI KT6 Delta).
HOWEVER, as many ACPI people already said several times, IT MAY BE RISKY AND
you CAN severely DAMAGE YOUR DATA when experimenting with this. So it's just
on you, how do you decide.
  With regards, Pavel Troller

Re: Question about AthlonXP on nforce 2 : C2-C3 states

[Arthur Huillet]

> On Athlon boards, I'm successfully using athcool
> ( http://members.jcom.home.ne.jp/jacobi/linux/files/athcool-0.3.11.tar.gz ),
> it substantially decreases CPU temperature even with C1 due to enabling
> "HALT Command detection" bit on the NB. 

Yes, I've already done that and it does work very well. 

> HOWEVER, as many ACPI people already said several times, IT MAY BE RISKY AND
> you CAN severely DAMAGE YOUR DATA when experimenting with this.

I guess so, otherwise the BIOS wouldn't probably be blocking those features. 
Thanks for the help, I won't test C2-C3 because the temperature results in C1 are already
excellent.

Now to take a look at cpufreq-nforce2 and have fun with it :) 
Have a nice day.
-- 
Greetings, 
A.H.