What To Do If Still Seeing Poor Linux Battery Life

Even after using the Linux 2.6.38 kernel power regression workaround for those systems affected by the ASPM (Active-State Power Management) bug that was detected by Phoronix, some Linux users still don't have their netbook/notebook battery lasting as long as they'd wish. For some, Microsoft Windows 7 is still performing better on the battery and/or Ubuntu 10.04 LTS and earlier. As I've said, there's still at least a couple more Linux power regressions and other areas for optimization that I've discovered and still haven't yet publicly documented in full, some in part due to still analyzing what's happening.

Thanks to the good advertising outcome (plus a pleasant increase in Phoronix Premium subscriptions and PayPal tips) from the power articles this week (the 2.6.38 regression and the Windows 7 vs. Ubuntu 11.04 power comparison) and some pleasurable advertising campaigns coming up in July (including a UPS power vendor), more of these Linux power items will be dug into next month and detailed on Phoronix. On my Twitter feed, I've been hinting at some things.

For those Linux users still reporting a large difference in power consumption when using Ubuntu 11.04 or any other modern Linux distribution compared to past Linux releases or Microsoft Windows, more information would be helpful. A list of notebooks/configurations where the battery life is still frustrating would be helpful for my purposes and I'm sure the community at large would like this data too.

Those that can definitively say their battery life is shorter on Linux currently compared to a past release or when running Windows, add your system information to this forum thread. At least say the notebook/netbook model number and the CPU/GPU. Ideally, install the Phoronix Test Suite (available on most distributions) and then run "phoronix-test-suite system-info" and report that output. Example system-info output:

Phoronix Test Suite v3.2.1
System Information

Hardware:
Processor: Intel Core i5 M 520 @ 2.39GHz (4 Cores), Motherboard: Intel 440BX, Chipset: Intel 440BX/ZX/DX, Memory: 2816MB, Disk: 19GB VMware Virtual I, Graphics: VMware SVGA II

Software:
OS: Ubuntu 10.10, Kernel: 2.6.36-020636-generic (x86_64), Desktop: GNOME 2.32.0, Display Server: X Server 1.9.0, Display Driver: vmware 11.0.1, OpenGL: 2.1 Mesa 7.9-devel, Compiler: GCC 4.4.5, File-System: ext4, Screen Resolution: 1920x963

That output would be most useful for full and accurate details on the system. The more information the better. It also doesn't matter if you're using Ubuntu, Fedora, Gentoo, or some other distribution, but as long as you can say with confident that the battery life is being degraded compared to the past or in relation to Microsoft Windows.

If you really want to make it worthwhile, run phoronix-test-suite benchmark battery-power-usage while on your notebook's battery. This will show the battery power usage when the system is idling, when the display is off, and then when the display is re-enabled and a short video demo is played.

Those numbers will at least provide some base reference for the kind of current power consumption to expect from the system. If you run this test, be sure to upload the results to OpenBenchmarking.org and link to the results from the aforementioned forum thread. Alternatively, with any Phoronix Test Suite test profile you can set MONITOR=sys.power as an environmental variable and then run any test profile to see the power usage while running the supplied workload (e.g. MONITOR=sys.power phoronix-test-suite benchmark nexuiz) as that's how all of the automated Phoronix benchmarking is carried out in base form.

If there's any clear trends of particular notebooks still having poor battery life with the latest Linux code, I (Phoronix Media) will likely purchase the same or similar notebooks (again, thanks go out to the Phoronix.com advertisers and Premium users, so please no AdBlock if this mission is to be a success) in an attempt to find even more Linux power regressions. Why? It seems no other distribution vendors or other stakeholders are doing anything significant about these Linux power issues, yet end-users are being disgruntled and complaining en mass. So many Linux distributions have been focusing on speeding up the boot process and other "fluffy" items, but it seems very few are interested in actually making Linux the most power efficient operating system; this would make a real difference. Some of the areas for Linux power optimization would also affect servers in the data center.

As I said in the earlier article, it takes about as much time to hit kernel regressions these days as it takes me to drink a Maß of Augustiner at Oktoberfest (not much time at all). This is done using the Phoronix Test Suite and related Phoronix components in a very easy and automated manner, so finding more regressions just does more to show off the software's capabilities in the enterprise world. If no vendors or developers step up to catch them first, it will be my personal vendetta and the results detailed on Phoronix. There's also some other features in the power management area I hope to work on soon for OpenBenchmarking.org.

So shall Linux power management be improved?