Tweaks To Extend The Battery Life Of Intel Linux Notebooks
The various kernel options were tested in an automated manner using the Phoronix Test Suite and the MATISK module for easy context-driven benchmarking.
Besides testing each kernel option alone, tests were also done with all options enabled and then also when forcing the PCI Express Active-State Power Management functionality. Fortunately, none of these kernel options had left the Hewlett-Packard EliteBook Sandy Bridge notebook being unstable or with any other apparent issues.
Tests profiles used included the battery-power-usage (idling, DPMS screen blanking, and video playback), OpenArena, and Nexuiz OpenGL benchmarks.
First is a look at the results for the Battery Power Usage test of idling at the Unity desktop, idling while the screen is powered off, and then when the screen is re-enabled with a short video being played back via MPlayer. Enabling RC6 was the biggest benefit with the power consumption dropping by 26% compared to the stock Linux 3.1 kernel. Enabling frame-buffer compression was not too useful for this test, down-clocking the LVDS dropped the power consumption by about 5%, and the already well tested PCI Express Active-State Power Management forcing dropped the power consumption by about 9%. When all of these options were enabled, the Sandy Bridge notebook power consumption went from an average of 17.75 Watts to 13.05 Watts, or a reduction of about 27% when all options were enabled together.