Re: [Announce] kernel-power stable v50 in Extras-Testing
Quote:
Originally Posted by Estel
(Post 1200881)
Any way to determine it for sure? Checking flags via ssh?
|
A simple way to test would be to run dbus-monitor on the system us. As you move the n900 through various orientations you will get messages on the bus WRT the phones position.
I have disabled the accelerometer by removing its module from mce.ini. This means no desktop rotation under any circumstance (unless a program specifically uses the accelerometer i.e. cloudgps)
It makes 0 difference to battery life. The 'heartbeat' daemon still runs and checks stuff anyway.
125MHz is pointless. The performance hit is so bad that the n900 lags like shít and the battery savings are negligible. The n900 sits at 125MHz for a short time, then as soon as there is any load too much it jumps to your maximum speed. Thus it spends most of the time at your max speed because 125 is so weak. The ONLY way you can use 125 is if you use the conservative governer with up_threshold 77, down threshold 75. This causes a LOT of frequency switching however the device does spend a lot of time at 125/250 while still being able to scale up effectively. It still lags like crap though.
Code:
#!/bin/sh
echo conservative > /opt/cpufreq/scaling_governor
echo 249600 > /opt/cpufreq/scaling_min_freq
echo 1500000 > /opt/cpufreq/conservative/sampling_rate
echo 11 > /opt/cpufreq/conservative/freq_step
echo "77" > /opt/cpufreq/conservative/up_threshold
echo "75" > /opt/cpufreq/conservative/down_threshold
I experimented with this for a while but could not find any benefit. (I also tried without the 250MHz minimum).
|