Is Linux Power Management Getting Better Or Worse?
With three laptops representing different generations of mobile hardware, we loaded up the past four stable releases of Fedora Linux plus the most recent Fedora 14 Alpha release and then carried out an arsenal of tests looking at how the battery power consumption rate has changed since 2008. If you are concerned at all about running Linux on your battery-powered mobile devices, this article is worth reading.
Fedora was used as our battery-testing platform due to its releases always living on the bleeding edge of free software development and plenty of upstream work occurring first in this Linux distribution. We tested every major release of Fedora from Fedora 10 "Cambridge" through the most recent Fedora 14 "Laughlin" Alpha build. We did not go further back than Fedora 10 in this article due to compatibility issues with some of the laptops. In a future article, we will also be testing each major kernel release to look at its battery/power performance rather than distribution releases comprising of updated desktop environments and other parts of the Linux software stack.
Fedora 10 "Leonidas" shipped with the Linux 2.6.27 kernel, GNOME 2.24.1, X.Org Server 1.5.3, xf86-video-nv 2.1.12, xf86-video-ati 6.9.0, Mesa 7.3-devel, GCC 4.3.2, and an EXT3 file-system. Fedora 11 "Leonidas" had the Linux 2.6.29 kernel, GNOME 2.26.1, X.Org Server 1.6.2 RC1, xf86-video-nouveau 0.0.10, xf86-video-ati 6.12.2, Mesa 7.5-devel, GCC 4.4.1, and an EXT4 file-system. Fedora 12 was codenamed Constantine and it offered up the Linux 2.6.31 kernel, GNOME 2.28.1, X.Org Server 1.7.1, xf86-video-nouveau 0.0.10, xf86-video-ati 6.12.99, Mesa 7.7-devel, GCC 4.4.4, and an EXT4 file-system. Next up was Goddard, or better known as Fedora 13, where there was the Linux 2.6.33 kernel, GNOME 2.30.0, X.Org Server 1.8.0, xf86-video-nouveau 0.0.15, xf86-video-ati 6.13.0, Mesa 7.8.1, GCC 4.4.4, and an EXT4 file-system. Lastly there was the Fedora 14 "Laughlin" Alpha release from last week with the Linux 2.6.35 kernel, GNOME 2.31.2, X.Org Server 1.9.0 RC5, xf86-video-nouveau 0.0.16, xf86-video-ati 6.13.99, Mesa 7.9-devel, GCC 4.5.1, and an EXT4 file-system. As only one of the notebooks had a 64-bit Intel CPU, for this testing today we used the 32-bit versions of Fedora across all three notebooks. Each Fedora release was done with a clean LiveCD installation and using the default settings.
Up for testing was a Lenovo ThinkPad R52, Lenovo ThinkPad T60, and a Lenovo ThinkPad T61. We had also hoped to use a netbook or two in this article, but due to a combination of Live USB installation problems with earlier releases and the USB DVD drive being used not working with the Fedora Live CD due to a missing driver, those had to be left out. Below are the three systems used for this power testing.
Lenovo ThinkPad R52
Processor: Intel Pentium M 1.86GHz @ 0.80GHz (Total Cores: 1), Motherboard: IBM 18494WU, Chipset: Intel Mobile 915GM/PM/GMS/910GML + ICH6M, Memory: 2004MB, Disk: 80GB Hitachi HTS541080G9AT00, Graphics: Mobility Radeon X300 (300/230MHz)
Lenovo ThinkPad T60
Processor: Intel CPU T2400 @ 1.83GHz (Total Cores: 2), Motherboard: LENOVO 2613EJU, Chipset: Intel Mobile 945GM/PM/GMS + ICH7-M, Memory: 1GB, Disk: 80GB Hitachi HTS541080G9SA00, Graphics: ATI Radeon Mobility X1400 (392/350MHz)
Lenovo ThinkPad T61
Processor: Intel Core 2 Duo CPU T9300 @ 2.50GHz (Total Cores: 2), Motherboard: LENOVO 6459CTO, Chipset: Intel Mobile PM965/GM965/GL960 + ICH8M-E, Memory: 3GB, Disk: 100GB Hitachi HTS72201, Graphics: nVidia Quadro NVS 140M
On these three laptops and five Fedora releases, we measured each of the battery's power consumption rate when the notebook was idling and then the display turned off via DPMS, while running the GraphicsMagick image manipulation program, while executing the PostMark disk benchmark, during media encoding with the x264 application, and while running the OpenArena game. During each of these test cases, the Phoronix Test Suite had recorded the power consumption rate via ACPI every second. Normally we publish this power data in the form of a line graph, but due to the mass amount of data being presented in this article, they were converted to bar graphs by averaging this power data.