After recently tracking down the major Linux kernel power regression that's present for a vast number of mobile users in Fedora 15, Ubuntu 11.04, and other recent Linux distributions shipping the 2.6.38+ kernel, the sights were turned to see how the power management of Ubuntu 11.04 compares to that of Windows 7 Professional Service Pack 1. In this article are the power consumption results of Ubuntu 11.04 compared directly to Windows 7 Professional 1 on several different systems with distinct notebook and desktop / workstation configurations.
For all testing, the Watts Up Pro power meter was used with its USB interfacing to the Phoronix Test Suite software for automatic monitoring. This monitoring was done by an independent system the entire time to ensure that there weren't any Windows/Linux software monitoring differences as the power meter was attached to each system's power supply.
The systems that where we looked at the Ubuntu 11.04 x86_64 vs. Windows 7 Professional Service Pack 1 x64 included:
Lenovo ThinkPad T61: Intel Core 2 Duo T9300, Hitachi 100GB HTS72201 SATA HDD, Intel PM965 + ICH8M-E, 4GB DDR3, NVIDIA Quadro NVS 140M 512MB.
Gulftown: Intel Core i7 990X, ASRock X58 SuperComputer, 3GB DDR3, 320GB Seagate ST3320620AS SATA HDD, NVIDIA GeForce 9800GTX.
Dual Opteron: 2 x AMD Opteron 2384 Quad-Core CPUs, Tyan S2932 motherboard, 160GB Western Digital WD1600YS-01S SATA HDD, 4GB ECC Registered DDR2, ATI FirePro V8700.
Phenom II: AMD Phenom II X3 710, MSI 890GXM-G65, Seagate 250GB ST250310AS SATA HDD, 4GB DDR3, AMD Radeon HD 4650.
A distinct selection of systems was sought after and the number of systems was just limited to time available. As only having a 64-bit edition of Windows 7 Professional, the hardware was also limited to newer platforms and no vintage hardware. Each operating system was cleanly installed and left in its configuration with stock settings. The respective proprietary graphics driver for each operating system was installed. As yesterday's article presented, the open-source GPU drivers do very bad at power management, besides their performance lacking.
The key areas we looked at were:
Boot / Start-Up: For five minutes the power consumption was polled on each system as it performed a cold boot followed by idling at the desktop (automatic log-in) for the remainder of the time.
Basic Desktop Usage: The power consumption from idling at the desktop to launching the respective web-browser (for Windows 7 SP1 that's Internet Explorer 9 and under Ubuntu 11.04 the default is Mozilla Firefox 4.0), navigating to Phoronix.com, and then launching the operating system's file manager, navigating each system's menus, etc. Just light desktop usage under each operating system.
Video Playback: The power consumption while watching the Flash-based HD version of Unigine's OilRush trailer on YouTube full-screen.
Gaming: The power consumption while running the open-source Nexuiz OpenGL game natively under each operating system.
Enough said, let's see the results.