From 1579100c872e6b76f09bd8f79dd85faf6076e5d9 Mon Sep 17 00:00:00 2001 From: Andrew Dodd Date: Sat, 15 Sep 2012 11:55:36 -0400 Subject: power_profile: Add cpu.awake parameter One of the remaining issues with the new profile is lack of a cpu.awake parameter - without this, wakelock holders won't be penalized in battery stats. Make a wild-ass guesstimate of 15 mA, based on: Best-case consumption is around 0.5%/hour with radio on, assume 8.25 mA consumption of non-CPU stuff in deep sleep. Best case consumption when wakelocked is usually around 1.25-1.5%/hour, assume then, about 25 mA total when wakelocked. Subtract the baseline and it's 15-16 mA. Let's just round to a multiple of 5 here, since it IS a WAG. Also, consumption when the CPU is running at 200 MHz (for example, governor set to ignore nice load, and a program is endlessly looping but niced to avoid bumping the gov) is usually around 4-5%/hour, which is semi-consistent with a number of 55 mA for 200 MHz op. Change-Id: I6aa08842dd2acfced0c70a95d973aa0b415799cc --- overlay/frameworks/base/core/res/res/xml/power_profile.xml | 1 + 1 file changed, 1 insertion(+) (limited to 'overlay') diff --git a/overlay/frameworks/base/core/res/res/xml/power_profile.xml b/overlay/frameworks/base/core/res/res/xml/power_profile.xml index 79996f7..ab187b6 100644 --- a/overlay/frameworks/base/core/res/res/xml/power_profile.xml +++ b/overlay/frameworks/base/core/res/res/xml/power_profile.xml @@ -25,6 +25,7 @@ 200000 2 + 15 577 408 -- cgit v1.1