Difference between revisions of "Talk:Problem with high power drain in ACPI sleep"

From ThinkWiki
Jump to: navigation, search
Line 21: Line 21:
  
 
-- Andrzej
 
-- Andrzej
 +
 +
Okay everyone. 
 +
 +
First off,  I wanted to make a comment about the "problem description" of this wiki page, which states, "Several people realized that their ThinkPads eat up too much power while suspended to ram via ACPI.  Compared to APM suspend to ram the power drain is experienced to be about 10 times as high, 2-5 Watts. This empties the battery within one or two days."  The admittedly optimistic capacity reported by my Thinkpad T42's <tt>/proc/acpi/battery/BAT0/info</tt> indicates that my battery can hold around 47520 mWh, which will last about 1 day at 2 Watts, and about 0.4 days at 5 Watts.  The original poster of the above problem description must have had a battery with twice the capacity.  I just wanted to throw that out there for anyone who thought that the ability to sleep 1 or 2 days didn't sound like it was 10 times worse than APM.
 +
 +
Secondly, I'm wondering what kind of power usage we're really shooting for with the ACPI S3 sleep state.  My system right now is eating up around 0.54 Watts when in the S3 state.  The <tt>/var/log/battery.log</tt> file created by the [[ACPI sleep power drain test script]] reports "Congratulations, your model seems NOT to be affected" if the power usage is merely below 1.0 Watt.  1.0 Watt seems like a very liberal threshold.  Can we get down to 0.2 Watts (which someone might think is possible from the problem description <math>-</math> a tenth of 2 Watts)?  Or is 0.5 Watts about as good as it gets?  I'm definately interested in everyone's experiences, opinions, and statistics.
 +
 +
-- Sukant

Revision as of 21:22, 21 July 2005

The explanation of the radeonfb fix is getting a bit disorganized; I defer to VB's expertise for the proper unified formulation. The points that should be made are:

  • It is important that the "radeonfb" kernel module gets loaded, even if you don't expect to use the console framebuffer. The default in FC4 (and probably other distros as well) is to just use the standard text-mode VGA driver for the console, and leave graphics to the X server; but the X.org "radeon" driver does not understand power management.
  • The radeonfb module can be loaded either at boot time, using "video=radeonfb", or at some later point (but before starting X) using "modprobe radeonfb".
  • Regardless of how the module is loaded, the parameter "radeon_force_sleep=1" enables the deep-sleep mode on chipset combinations that have not been explicitly whitelisted yet.

-- Andrzej (2005-07-02)


I agree. Loading "radeonfb" is not FC specific. It should be stated in more general. Question is if "radeon_force_sleep=1" is FC specific. If it is so that has to be stated, if not, it should be written in a more general way. However, VB wrote, the fix is not in official kernels, hence i guess it's FC4 specific. VB, will you do the job?

Wyrfel.


I made some changes to address your suggestions. Better?

Volker


Looks great to me!

-- Andrzej

Okay everyone.

First off, I wanted to make a comment about the "problem description" of this wiki page, which states, "Several people realized that their ThinkPads eat up too much power while suspended to ram via ACPI. Compared to APM suspend to ram the power drain is experienced to be about 10 times as high, 2-5 Watts. This empties the battery within one or two days." The admittedly optimistic capacity reported by my Thinkpad T42's /proc/acpi/battery/BAT0/info indicates that my battery can hold around 47520 mWh, which will last about 1 day at 2 Watts, and about 0.4 days at 5 Watts. The original poster of the above problem description must have had a battery with twice the capacity. I just wanted to throw that out there for anyone who thought that the ability to sleep 1 or 2 days didn't sound like it was 10 times worse than APM.

Secondly, I'm wondering what kind of power usage we're really shooting for with the ACPI S3 sleep state. My system right now is eating up around 0.54 Watts when in the S3 state. The /var/log/battery.log file created by the ACPI sleep power drain test script reports "Congratulations, your model seems NOT to be affected" if the power usage is merely below 1.0 Watt. 1.0 Watt seems like a very liberal threshold. Can we get down to 0.2 Watts (which someone might think is possible from the problem description <math>-</math> a tenth of 2 Watts)? Or is 0.5 Watts about as good as it gets? I'm definately interested in everyone's experiences, opinions, and statistics.

-- Sukant