Skip to content

Official sunxi OpenEmbedded layer for Allwinner-based boards.

License

Notifications You must be signed in to change notification settings

vishnupatekar/meta-sunxi

 
 

Repository files navigation

meta-sunxi

Official sunxi OpenEmbedded layer for Allwinner-based boards.

This layer depends on the additional layer:

meta-openembedded: git://git.openembedded.org/meta-openembedded

Tested with core-image-base.

Maintainers:

Performance

The default machine settings are meant to be the lowest common denominator, maximizing generality. Significantly better performance (2x-3x) can be achieved with the following settings:

Allwinner A20

For Allwinner A20 (Cubieboard2/CubieTruck), the following tuning options are recommended:

Enable hardfloat, thumb2 and neon capabilities

DEFAULTTUNE = "cortexa7hf-neon-vfpv4"

This tuning profile takes advantage of the Allwinner A20 hardfloat, neon and vfpv4 capabilities.

Change CPU governor to ondemand, and tune settings

echo ondemand > /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor
echo 336000 > /sys/devices/system/cpu/cpu0/cpufreq/scaling_min_freq
echo 912000 > /sys/devices/system/cpu/cpu0/cpufreq/scaling_max_freq
#More aggressive
#echo 1008000 > /sys/devices/system/cpu/cpu0/cpufreq/scaling_max_freq
echo 40 > /sys/devices/system/cpu/cpufreq/ondemand/up_threshold
echo 200000 > /sys/devices/system/cpu/cpufreq/ondemand/sampling_rate

This code changes the default CPU governor from fantasy to ondemand, and tunes its settings, as recommended at http://linux-sunxi.org/Cpufreq

For additional discussion, see linux-sunxi#25

About

Official sunxi OpenEmbedded layer for Allwinner-based boards.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • C 65.2%
  • BitBake 34.8%