battery life with cpufreq

battery life with cpufreq

[Fred Blaise]

Hello all

I have an dell inspiron 5150, 3Ghz. I have loaded p4_clockmod, and my
CPU frequency now changes with powernowd. It seems to work well.

`--> cat /proc/cpufreq 
          minimum CPU frequency  -  maximum CPU frequency  -  policy
CPU  0       383323 kHz ( 12 %)  -    3066590 kHz (100 %)  -  userspace

`--> cat throttling 
state count:             8
active state:            T0
states:
   *T0:                  00%
    T1:                  12%
    T2:                  25%
    T3:                  37%
    T4:                  50%
    T5:                  62%
    T6:                  75%
    T7:                  87%

However, I am not seeing any battery life improvement.

Any ideas?

Thanks

fred


-------------------------------------------------------
SF.Net email is sponsored by:
Tame your development challenges with Apache's Geronimo App Server.
Download it for free - -and be entered to win a 42" plasma tv or your very
own Sony(tm)PSP.  Click here to play: http://sourceforge.net/geronimo.php

Re: battery life with cpufreq

[Dominik Brodowski]

Hi,

On Mon, Sep 19, 2005 at 11:11:01AM +0200, Fred Blaise wrote:
> I have an dell inspiron 5150, 3Ghz. I have loaded p4_clockmod, and my
> CPU frequency now changes with powernowd. It seems to work well.
>
> However, I am not seeing any battery life improvement.
> 
> Any ideas?

Yes. In general, throttling doesn't improve battery life, only CPU frequency
scaling does:

"In contrast to CPU frequency scaling, where the operating frequency is
constantly modulated, throttling means the CPU is forced to a halt for short
periods of time. If throttled, the CPU [enters] into a physical and
electrical state comparable to the idling states mentioned above, so it can
be described as an enforced idling of the CPU.

As certain CPU power state typically utilize similar hardware
implementations, and as throttling does not have a positive effect on the
energy consumption during CPU power states, throttling the CPU by a given
rate is only useful if the CPU is less idle than the throttling rate."

	Dominik


-------------------------------------------------------
SF.Net email is sponsored by:
Tame your development challenges with Apache's Geronimo App Server. 
Download it for free - -and be entered to win a 42" plasma tv or your very
own Sony(tm)PSP.  Click here to play: http://sourceforge.net/geronimo.php

Re: battery life with cpufreq

[Fred Blaise]

On 9/19/05, Dominik Brodowski <linux-X3ehHDuj6sIIGcDfoQAp7OTW4wlIGRCZ@public.gmane.org> wrote:
> Hi,
> 
> On Mon, Sep 19, 2005 at 11:11:01AM +0200, Fred Blaise wrote:
> > I have an dell inspiron 5150, 3Ghz. I have loaded p4_clockmod, and my
> > CPU frequency now changes with powernowd. It seems to work well.
> >
> > However, I am not seeing any battery life improvement.
> >
> > Any ideas?
> 
> Yes. In general, throttling doesn't improve battery life, only CPU frequency
> scaling does:
ok.. So the fact that my CPU goes down to 383Mghz instead of the
normal 3Ghz is not frequency scaling? (pardon the noob that I am if
this question doesn't make sense)
> 
> "In contrast to CPU frequency scaling, where the operating frequency is
> constantly modulated, throttling means the CPU is forced to a halt for short
> periods of time. If throttled, the CPU [enters] into a physical and
> electrical state comparable to the idling states mentioned above, so it can
> be described as an enforced idling of the CPU.
> 
> As certain CPU power state typically utilize similar hardware
> implementations, and as throttling does not have a positive effect on the
> energy consumption during CPU power states, throttling the CPU by a given
> rate is only useful if the CPU is less idle than the throttling rate."
> 
>         Dominik
>


-------------------------------------------------------
SF.Net email is sponsored by:
Tame your development challenges with Apache's Geronimo App Server.
Download it for free - -and be entered to win a 42" plasma tv or your very
own Sony(tm)PSP.  Click here to play: http://sourceforge.net/geronimo.php

Re: battery life with cpufreq

[Timo Hoenig]

Hi,

On Mon, 2005-09-19 at 11:35 +0200, Fred Blaise wrote:

[...]

> ok.. So the fact that my CPU goes down to 383Mghz instead of the
> normal 3Ghz is not frequency scaling? (pardon the noob that I am if
> this question doesn't make sense)

This is frequency scaling.  But it has nothing in common with
throttling.

Are you saying that you're stepping down from 3 GHz to 300 MHz and do
not see any improvements regarding battery life time?

[...]

See you,

   Timo



-------------------------------------------------------
SF.Net email is sponsored by:
Tame your development challenges with Apache's Geronimo App Server. 
Download it for free - -and be entered to win a 42" plasma tv or your very
own Sony(tm)PSP.  Click here to play: http://sourceforge.net/geronimo.php

Re: battery life with cpufreq

[Fred Blaise]

On 9/19/05, Timo Hoenig <thoenig-dCxI//HcOdFeoWH0uzbU5w@public.gmane.org> wrote:
> Hi,
> 
> On Mon, 2005-09-19 at 11:35 +0200, Fred Blaise wrote:
> 
> [...]
> 
> > ok.. So the fact that my CPU goes down to 383Mghz instead of the
> > normal 3Ghz is not frequency scaling? (pardon the noob that I am if
> > this question doesn't make sense)
> 
> This is frequency scaling.  But it has nothing in common with
> throttling.
> 
> Are you saying that you're stepping down from 3 GHz to 300 MHz and do
> not see any improvements regarding battery life time?
Yes. 
`--> cat cpuinfo 
processor       : 0
vendor_id       : GenuineIntel
cpu family      : 15
model           : 2
model name      : Mobile Intel(R) Pentium(R) 4     CPU 3.06GHz
stepping        : 9
cpu MHz         : 382.040
[...]
> 
> [...]
> 
> See you,
> 
>    Timo
> 
>


-------------------------------------------------------
SF.Net email is sponsored by:
Tame your development challenges with Apache's Geronimo App Server.
Download it for free - -and be entered to win a 42" plasma tv or your very
own Sony(tm)PSP.  Click here to play: http://sourceforge.net/geronimo.php

Re: battery life with cpufreq

[Dominik Brodowski]

Hi,

On Mon, Sep 19, 2005 at 11:45:38AM +0200, Timo Hoenig wrote:
> > ok.. So the fact that my CPU goes down to 383Mghz instead of the
> > normal 3Ghz is not frequency scaling? (pardon the noob that I am if
> > this question doesn't make sense)
> 
> This is frequency scaling.  But it has nothing in common with
> throttling.

No, p4-clockmod is frequency throttling or frequency modulation, but definitely 
not scaling.

	Dominik


-------------------------------------------------------
SF.Net email is sponsored by:
Tame your development challenges with Apache's Geronimo App Server. 
Download it for free - -and be entered to win a 42" plasma tv or your very
own Sony(tm)PSP.  Click here to play: http://sourceforge.net/geronimo.php

Re: battery life with cpufreq

[Fred Blaise]

On 9/19/05, Dominik Brodowski <linux-X3ehHDuj6sIIGcDfoQAp7OTW4wlIGRCZ@public.gmane.org> wrote:
> Hi,
> 
> On Mon, Sep 19, 2005 at 11:45:38AM +0200, Timo Hoenig wrote:
> > > ok.. So the fact that my CPU goes down to 383Mghz instead of the
> > > normal 3Ghz is not frequency scaling? (pardon the noob that I am if
> > > this question doesn't make sense)
> >
> > This is frequency scaling.  But it has nothing in common with
> > throttling.
> 
> No, p4-clockmod is frequency throttling or frequency modulation, but definitely
> not scaling.

Ok.. well.. speedstep* do not work here (my frequency will remain
stuck where it last was). On the other hand, p4_clockmod, i can see in
cpuinfo the frequency going up and down.

Here are the relevant modules loaded:
`--> lsmod 
[...]     
speedstep_lib           3940  1 p4_clockmod
cpufreq_powersave       1632  0 
cpufreq_userspace       4348  0 
cpufreq_ondemand        6140  0
freq_table              4004  1 p4_clockmod
[...]

So you would be saying that even though I can see the frequency change
in cpuinfo, it is kind of "lying" to me? I can feel the difference
too, when I use the laptop. It definitly has the speed of a 383Mghz
when the cpuinfo shows it at that frequency..
Thanks
>         Dominik
fred
>


-------------------------------------------------------
SF.Net email is sponsored by:
Tame your development challenges with Apache's Geronimo App Server.
Download it for free - -and be entered to win a 42" plasma tv or your very
own Sony(tm)PSP.  Click here to play: http://sourceforge.net/geronimo.php

Re: battery life with cpufreq

[Dominik Brodowski]

Hi,

On Mon, Sep 19, 2005 at 11:53:34AM +0200, Fred Blaise wrote:
> Ok.. well.. speedstep* do not work here (my frequency will remain
> stuck where it last was). On the other hand, p4_clockmod, i can see in
> cpuinfo the frequency going up and down.

Yes, the "frequency" which is reported here is how it "feels", kind of a
normalized variant of 
	x x x x x x x x x x x x x x x x x x x x x x x x 
	x x x x         x x x x       x x x x        
	------------------------------------------------>t

As these "stops" don't last for long periods of time, it is quite correct to
say that the CPU only executes half as much instructions during one second,
so it runs at "half frequency".

	Dominik


-------------------------------------------------------
SF.Net email is sponsored by:
Tame your development challenges with Apache's Geronimo App Server. 
Download it for free - -and be entered to win a 42" plasma tv or your very
own Sony(tm)PSP.  Click here to play: http://sourceforge.net/geronimo.php

Re: battery life with cpufreq

[Pavel Troller]

> Yes, the "frequency" which is reported here is how it "feels", kind of a
> normalized variant of 
> 	x x x x x x x x x x x x x x x x x x x x x x x x 
> 	x x x x         x x x x       x x x x        
> 	------------------------------------------------>t
> 
> As these "stops" don't last for long periods of time, it is quite correct to
> say that the CPU only executes half as much instructions during one second,
> so it runs at "half frequency".
> 
Hi,
  ok, it's clear. But isn't it true that in the periods with clock stopped
the CPU consumes much less power than in those with clock running ? Why there
is no at least slight power saving effect then? 
  BTW I'm experiencing the same. p4-clockmod also doesn't save power on my
Clevo D610SU notebook.
                                  With regards, Pavel Troller


-------------------------------------------------------
SF.Net email is sponsored by:
Tame your development challenges with Apache's Geronimo App Server. 
Download it for free - -and be entered to win a 42" plasma tv or your very
own Sony(tm)PSP.  Click here to play: http://sourceforge.net/geronimo.php

Re: battery life with cpufreq

[Dominik Brodowski]

> Hi,
>   ok, it's clear. But isn't it true that in the periods with clock stopped
> the CPU consumes much less power than in those with clock running ?

Yes, indeed. However, it consumes exactly the same amount of energy as if it
were in an "idle state" (ACPI C-States). And that's entered automatically if
there is nothing for the CPU to do. As a typical CPU in a typical notebook
only has something to do for 5% of the time, throttling to 75%, for example,
means the CPU is:
	working		 5%
	idling		20%
	throttling	75%

moving "time" around from idling to throttling or the other way around
doesn't matter anything. If you use CPU frequency scaling, though, the
"working" part consumes less CPU power. Also, the "idle state" typically
consumes less power if the CPU frequency and voltage was scaled down.

	Dominik


-------------------------------------------------------
SF.Net email is sponsored by:
Tame your development challenges with Apache's Geronimo App Server. 
Download it for free - -and be entered to win a 42" plasma tv or your very
own Sony(tm)PSP.  Click here to play: http://sourceforge.net/geronimo.php

Re: battery life with cpufreq

[Fred Blaise]

On 9/19/05, Dominik Brodowski <linux-X3ehHDuj6sIIGcDfoQAp7OTW4wlIGRCZ@public.gmane.org> wrote:
> > Hi,
> >   ok, it's clear. But isn't it true that in the periods with clock stopped
> > the CPU consumes much less power than in those with clock running ?
> 
> Yes, indeed. However, it consumes exactly the same amount of energy as if it
> were in an "idle state" (ACPI C-States). And that's entered automatically if
> there is nothing for the CPU to do. As a typical CPU in a typical notebook
> only has something to do for 5% of the time, throttling to 75%, for example,
> means the CPU is:
>         working          5%
>         idling          20%
>         throttling      75%
> 
> moving "time" around from idling to throttling or the other way around
> doesn't matter anything. If you use CPU frequency scaling, though, the
> "working" part consumes less CPU power. Also, the "idle state" typically
> consumes less power if the CPU frequency and voltage was scaled down.
To sum up, I am screwed I guess :\ Since the speedstep* don't work for
me... No long battery life for me...

Thanks for your explanations.
> 
>         Dominik
fred
>


-------------------------------------------------------
SF.Net email is sponsored by:
Tame your development challenges with Apache's Geronimo App Server.
Download it for free - -and be entered to win a 42" plasma tv or your very
own Sony(tm)PSP.  Click here to play: http://sourceforge.net/geronimo.php

Re: battery life with cpufreq

[Erik Slagter]

[-- Attachment #1: Type: text/plain, Size: 510 bytes --]

On Mon, 2005-09-19 at 13:39 +0200, Fred Blaise wrote:
> > moving "time" around from idling to throttling or the other way around
> > doesn't matter anything. If you use CPU frequency scaling, though, the
> > "working" part consumes less CPU power. Also, the "idle state" typically
> > consumes less power if the CPU frequency and voltage was scaled down.
> To sum up, I am screwed I guess :\ Since the speedstep* don't work for
> me... No long battery life for me...

Or find out why it doesn't work.

[-- Attachment #2: smime.p7s --]
[-- Type: application/x-pkcs7-signature, Size: 2115 bytes --]

Re: battery life with cpufreq

[Johan Vromans]

Fred Blaise <chapeaurouge-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org> writes:

> To sum up, I am screwed I guess :\ Since the speedstep* don't work for
> me... No long battery life for me...

Centrino cpustep didn't work for me either, but acpi-cpufreq does. 
Have you tried it?

-- Johan


-------------------------------------------------------
SF.Net email is sponsored by:
Tame your development challenges with Apache's Geronimo App Server. 
Download it for free - -and be entered to win a 42" plasma tv or your very
own Sony(tm)PSP.  Click here to play: http://sourceforge.net/geronimo.php

Re: Re: battery life with cpufreq

[Fred Blaise]

On 9/19/05, Johan Vromans <jvromans-2pNSKKP3PSKEVqv0pETR8A@public.gmane.org> wrote:
> Fred Blaise <chapeaurouge-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org> writes:
> 
> > To sum up, I am screwed I guess :\ Since the speedstep* don't work for
> > me... No long battery life for me...
> 
> Centrino cpustep didn't work for me either, but acpi-cpufreq does.
> Have you tried it?
Just did...
FATAL: Error inserting acpi_cpufreq
(/lib/modules/2.6.10fb/kernel/arch/i386/kernel/cpu/cpufreq/acpi-cpufreq.ko):
No such device

I haven't done any research on that yet, so this will be for later.

Thanks all for the time and help.
> 
> -- Johan
fred
> 
> 
> -------------------------------------------------------
> SF.Net email is sponsored by:
> Tame your development challenges with Apache's Geronimo App Server.
> Download it for free - -and be entered to win a 42" plasma tv or your very
> own Sony(tm)PSP.  Click here to play: http://sourceforge.net/geronimo.php
> _______________________________________________
> Acpi-devel mailing list
> Acpi-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org
> https://lists.sourceforge.net/lists/listinfo/acpi-devel
>


-------------------------------------------------------
SF.Net email is sponsored by:
Tame your development challenges with Apache's Geronimo App Server.
Download it for free - -and be entered to win a 42" plasma tv or your very
own Sony(tm)PSP.  Click here to play: http://sourceforge.net/geronimo.php

Re: battery life with cpufreq

[Dominik Brodowski]

Hi,

On Mon, Sep 19, 2005 at 11:35:24AM +0200, Fred Blaise wrote:
> On 9/19/05, Dominik Brodowski <linux-X3ehHDuj6sIIGcDfoQAp7OTW4wlIGRCZ@public.gmane.org> wrote:
> > Hi,
> > 
> > On Mon, Sep 19, 2005 at 11:11:01AM +0200, Fred Blaise wrote:
> > > I have an dell inspiron 5150, 3Ghz. I have loaded p4_clockmod, and my
> > > CPU frequency now changes with powernowd. It seems to work well.
> > >
> > > However, I am not seeing any battery life improvement.
> > >
> > > Any ideas?
> > 
> > Yes. In general, throttling doesn't improve battery life, only CPU frequency
> > scaling does:
> ok.. So the fact that my CPU goes down to 383Mghz instead of the
> normal 3Ghz is not frequency scaling? (pardon the noob that I am if
> this question doesn't make sense)

Exactly. In fact, the frequency rate is not changed, it is only stopped for
short periods of time.



3 Ghz	  	  :	x x x x x x x x x x x x x x x x
scaling to 1.5 GHz:	x   x   x   x   x   x   x   x 
throttling to 1.5 :	x x x x       x x x x         x
			-------------------------------> t
			[x] denotes an instruction is executed by the CPU

If either the acpi-cpufreq or one of the speedstep-* modules works, you
should use that instead of p4-clockmod -- these offer (under normal
circumstances) CPU frequency scaling, while p4-clockmod only offers
throttling.

	Dominik


-------------------------------------------------------
SF.Net email is sponsored by:
Tame your development challenges with Apache's Geronimo App Server. 
Download it for free - -and be entered to win a 42" plasma tv or your very
own Sony(tm)PSP.  Click here to play: http://sourceforge.net/geronimo.php