Inspiron 6000 and CPU power saving

Inspiron 6000 and CPU power saving

[Michael (Micksa) Slade]

I recently discovered that my Inspiron 6000 uses about 50% more power 
idling in linux than in windows XP.  This means its battery life is 
about 2/3 of what it could/should be.

I guessed it might be the CPU, and did some tests.  The results strongly 
suggest as much.  These are the results I got for power consumption in 
various situations.

linux idle at 800MHz: 27W        
linux idle at 1600MHz: 36W        
linux raytracing at 800: 30W
linux raytracing at 1600: 42W 

windows idle (presumably 800MHz): 16W
windows raytracing (presumably 1600MHz): 36W

I've tried ubuntu dapper and ubuntu edgy, and RIP 10 (rescue disk) and 
BBC 2.1 (rescue disk), and they all appear to have the same issue.  The 
machine's BIOS has no APM so I can't try it for comparison.

I've tried noapic and "echo n > 
/sys/module/processor/parameters/max_cstate", where n is 1 thru 4.  
Neither appear to have any affect.

I need help digging deeper.  I guess /proc/acpi/processor/CPU0/power 
could give some insight but I'm not sure how to read the numbers.  That 
and "learn about ACPI" is all I can figure out so far.

So where to from here?  I am prepared to spend a significant amount of 
time researching and resolving the issue, so feel free to suggest 
reading the ACPI spec or whatever if that's what it's going to take.

Mick.

Re: Inspiron 6000 and CPU power saving

[Oleg Verych]

Hallo, Michael.

On 2006-10-16, Michael (Micksa) Slade wrote:
> I recently discovered that my Inspiron 6000 uses about 50% more power 
> idling in linux than in windows XP.  This means its battery life is 
> about 2/3 of what it could/should be.
>
> I guessed it might be the CPU, and did some tests.  The results strongly 
> suggest as much.  These are the results I got for power consumption in 
> various situations.
>
> linux idle at 800MHz: 27W        
> linux idle at 1600MHz: 36W        
> linux raytracing at 800: 30W
> linux raytracing at 1600: 42W 
>
> windows idle (presumably 800MHz): 16W
> windows raytracing (presumably 1600MHz): 36W
>
> I've tried ubuntu dapper and ubuntu edgy, and RIP 10 (rescue disk) and 
> BBC 2.1 (rescue disk), and they all appear to have the same issue.  The 
> machine's BIOS has no APM so I can't try it for comparison.

After reading all that, i think this is wrong list to post such things.

Anyway.
Note, that you didn't mentioned yours shiny laptop's architecture.
More to that, if it's cool-modern CPU from Intel or just a AMD's one,
what linux-kernel version you're running? As for me, i don't know
neither what Inspiron 6000 is, nor kernel versions "ubuntu dapper
(edgy)" is running. And if you will know all that information, better keep
it for your's google skills, not for further postings here.

Also if you want to compare power consumption, better include time
tasks ran, thus one can estimate real *work* done by system for
application and finally amount of actual heat.

> I need help digging deeper.  I guess /proc/acpi/processor/CPU0/power 
> could give some insight but I'm not sure how to read the numbers.  That 
> and "learn about ACPI" is all I can figure out so far.

Search lkml for 'ACPI' and 'Linus', after that, just relax and reboot to
windo$ and forget anything else ;)

> So where to from here?  I am prepared to spend a significant amount of 
> time researching and resolving the issue, so feel free to suggest 
> reading the ACPI spec or whatever if that's what it's going to take.

If you _really_ want to do something --  read sources, build kernels,
tune kernels and somewhere in between RTFM. And finally share results.
____

Re: Inspiron 6000 and CPU power saving

[Pavel Machek]

Hi!

> I recently discovered that my Inspiron 6000 uses about 50% more power 
> idling in linux than in windows XP.  This means its battery life is 
> about 2/3 of what it could/should be.
> 
> I guessed it might be the CPU, and did some tests.  The results strongly 
> suggest as much.  These are the results I got for power consumption in 
> various situations.
> 
> linux idle at 800MHz: 27W        
> linux idle at 1600MHz: 36W        
> linux raytracing at 800: 30W
> linux raytracing at 1600: 42W 
> 
> windows idle (presumably 800MHz): 16W
> windows raytracing (presumably 1600MHz): 36W
> 
> I've tried ubuntu dapper and ubuntu edgy, and RIP 10 (rescue disk) and 
> BBC 2.1 (rescue disk), and they all appear to have the same issue.  The 
> machine's BIOS has no APM so I can't try it for comparison.

Let me guess, UHCI? Remove usb modules, and measure again.

For details, see 8hours on your lap presentation, linked from
http://www.livejournal.com/~pavelmachek .

								Pavel

RE: Inspiron 6000 and CPU power saving

[Pallipadi, Venkatesh]

 

>-----Original Message-----
>From: linux-kernel-owner@vger.kernel.org 
>[mailto:linux-kernel-owner@vger.kernel.org] On Behalf Of 
>Michael (Micksa) Slade
>Sent: Sunday, October 15, 2006 7:18 PM
>To: linux-kernel@vger.kernel.org
>Subject: Inspiron 6000 and CPU power saving
>
>I recently discovered that my Inspiron 6000 uses about 50% more power 
>idling in linux than in windows XP.  This means its battery life is 
>about 2/3 of what it could/should be.
>
>I guessed it might be the CPU, and did some tests.  The 
>results strongly 
>suggest as much.  These are the results I got for power consumption in 
>various situations.
>
>linux idle at 800MHz: 27W        
>linux idle at 1600MHz: 36W        
>linux raytracing at 800: 30W
>linux raytracing at 1600: 42W 
>
>windows idle (presumably 800MHz): 16W
>windows raytracing (presumably 1600MHz): 36W
>
>I've tried ubuntu dapper and ubuntu edgy, and RIP 10 (rescue disk) and 
>BBC 2.1 (rescue disk), and they all appear to have the same 
>issue.  The 
>machine's BIOS has no APM so I can't try it for comparison.
>
>I've tried noapic and "echo n > 
>/sys/module/processor/parameters/max_cstate", where n is 1 thru 4.  
>Neither appear to have any affect.
>
>I need help digging deeper.  I guess /proc/acpi/processor/CPU0/power 
>could give some insight but I'm not sure how to read the 
>numbers.  That 
>and "learn about ACPI" is all I can figure out so far.
>
>So where to from here?  I am prepared to spend a significant amount of 
>time researching and resolving the issue, so feel free to suggest 
>reading the ACPI spec or whatever if that's what it's going to take.
>
>Mick.
>

Output of 
#cat /proc/acpi/processor/CPU0/power/*
And
#cat /sys/devices/system/cpu/cpu0/cpufreq/*
Will be a good starting point.

Also, open a issue at bugme.osdl.org. It makes tracking the issues
easier that way.

Thanks,
Venki

Re: Inspiron 6000 and CPU power saving

[Michael (Micksa) Slade]

Pallipadi, Venkatesh wrote:

> 
>
>  
>
>>-----Original Message-----
>>From: linux-kernel-owner@vger.kernel.org 
>>[mailto:linux-kernel-owner@vger.kernel.org] On Behalf Of 
>>Michael (Micksa) Slade
>>Sent: Sunday, October 15, 2006 7:18 PM
>>To: linux-kernel@vger.kernel.org
>>Subject: Inspiron 6000 and CPU power saving
>>
>>I recently discovered that my Inspiron 6000 uses about 50% more power 
>>idling in linux than in windows XP.  This means its battery life is 
>>about 2/3 of what it could/should be.
>>
>>I guessed it might be the CPU, and did some tests.  The 
>>results strongly 
>>suggest as much.  These are the results I got for power consumption in 
>>various situations.
>>
>>linux idle at 800MHz: 27W        
>>linux idle at 1600MHz: 36W        
>>linux raytracing at 800: 30W
>>linux raytracing at 1600: 42W 
>>
>>windows idle (presumably 800MHz): 16W
>>windows raytracing (presumably 1600MHz): 36W
>>
>>I've tried ubuntu dapper and ubuntu edgy, and RIP 10 (rescue disk) and 
>>BBC 2.1 (rescue disk), and they all appear to have the same 
>>issue.  The 
>>machine's BIOS has no APM so I can't try it for comparison.
>>
>>I've tried noapic and "echo n > 
>>/sys/module/processor/parameters/max_cstate", where n is 1 thru 4.  
>>Neither appear to have any affect.
>>
>>I need help digging deeper.  I guess /proc/acpi/processor/CPU0/power 
>>could give some insight but I'm not sure how to read the 
>>numbers.  That 
>>and "learn about ACPI" is all I can figure out so far.
>>
>>So where to from here?  I am prepared to spend a significant amount of 
>>time researching and resolving the issue, so feel free to suggest 
>>reading the ACPI spec or whatever if that's what it's going to take.
>>
>>Mick.
>>
>>    
>>
>
>Output of 
>#cat /proc/acpi/processor/CPU0/power/*
>And
>#cat /sys/devices/system/cpu/cpu0/cpufreq/*
>Will be a good starting point.
>
>Also, open a issue at bugme.osdl.org. It makes tracking the issues
>easier that way.
>
>Thanks,
>Venki
>  
>
Done.

http://bugzilla.kernel.org/show_bug.cgi?id=7393

As part of this I've also tried vanilla 2.6.19-rc2 and done another 
small test. See the report.

Mick.