SATA Aggressive Link Power Management (ALPM) - call for testing

Colin Ian King colin.king at
Tue Dec 6 20:44:30 UTC 2011

On 06/12/11 17:14, Tim Gardner wrote:
> On 12/06/2011 06:25 AM, Colin Ian King wrote:
>> Hi there,
>> Aggressive Link Power Management (ALPM) is a mechanism where a SATA AHCI
>> controller can put the SATA link that connects to the disk into a very
>> low power mode during periods of zero I/O activity and into an active
>> power state when work needs to be done. Tests show that this can save
>> around 0.5-1.5 Watts of power on a typical system.
>> ALPM is now available in several SATA controllers that use the Advanced
>> Host Controller Interface (AHCI). However, there is some anecdotal
>> evidence that some controllers may go into a low power state incorrectly
>> and this ends up causing data loss. Ubuntu has the ability to use ALPM
>> but it it disabled by default since it can cause data loss on some
>> machines.
>> We are looking for (brave?) volunteers to test ALPM for 2 reasons:
>> 1) help to identify typical power savings on a range of machines
>> 2) help to identify chipsets (and machines) where ALPM works reliably
>> and also where it is broken and needs fixing.
>> An ALPM crowd-sourcing testing Wiki page has been created that describes
>> a testing methodology and has a table for test results:
>> Please note that there is a possibility that ALPM MAY CAUSE DATA LOSS on
>> some machines, so please ensure you have backed up your data or don't
>> mind the risk in losing your data.
>> Colin
> How about screen backlight management? I've noticed that my screen at 
> full brightness consumes about 4W compared to screen off. Should we at 
> least be measuring that ?

measuring this is fairly easy but I'm not sure what you'd like tested.   
Backlights are adjusted on AC or battery automatically and users can 
tweak these using hotkeys too.  So, do we want to accurately know just 
how much power
we consume, with different levels?

I suspect I should also investigate something that reports:

"Interestingly, in our measurements we have found that, at least on 
newer laptops, the kernel sysfs method (which tends to use the BIOS on 
the machine) seems to not control the actual backlight, but rather 
change the colors of the pixels on the screen to appear darker. Using 
darker pixels obviously doesn't save as much power as reducing the 
intensity of the backlight." that I will at least poke at that.

The backlight issue makes me also think about other non-obvious 
settings. For example, with framebuffer compression we use less power to 
clock out the image, so I'd like to measure a default desktop background 
vs a plain background vs a complex background image.

And also I will be investigating the white vs black pixels setting when 
the screen is blanked - apparently white saves power.

If any one has any further ideas, let me know.


More information about the kernel-team mailing list