Creating Energy-Efficient Software (Part 3)

To save energy through software, we have described techniques for computational efficiency and for data efficiency. A third paradigm to address is context awareness. Context awareness was first introduced by Schilit in 1994 [Ref18]. The objective is to create applications that can respond or adapt to changes in the environment. For the physical environment this requires sensors and the ability to generate events or state changes to which the applications can react. For example, many notebook computers respond to the change from AC to battery power by automatically dimming the display. Some notebook PCs park the hard drive heads when sensors detect that the device is falling – to avoid a head crash. Some applications may write cached data to flash when the battery is getting critically low.

Context may also apply to a user’s situation. For example, a software application acting as an intelligent travel assistant may find and offer alternative flights when it’s apparent based on your location (stuck in traffic) that you have missed your scheduled departure.

With respect to power, applications should respond to system changes and take actions that will conserve energy – or at least offer options to the user. These events include AC vs. battery, charge remaining on the battery, and the state of various energy consuming devices such as WiFi and Bluetooth radios.

The next section describes techniques for creating more energy efficient games and a later section describes various tools that support the development of context aware applications.

PC gaming has traditionally been associated with desktop sys tems due to the performance requirements of game applications. As laptop capabilities become comparable to desktops, some gamers consider these as gaming platforms. Battery life is an important factor for system usability. To determine the most power efficient configuration, this section analyzes the power consumption of laptops while changing different game applications settings like graphics, resolution, FPS, and more. The study focuses on how the different game settings can provide a more efficient power configuration with a battery-powered laptop and provides recommendations on how to optimize games for power.[9]

All the tests described here were conducted on one of two configurations:

• A 2.0 GHz dual-core Intel Dual Core engineering system with an nVidia 7800 graphics card. Power measurements taken with a Fluke NetDAQ* - see Appendix A
• A Sony* VAIO, Centrino Core Duo 2.0GHz, with an nVidia 7400 graphics card. Power measurements taken with an Extech Power Analyzer[10]

Three off-the-shelf games were used during the measurements (2 first-person shooter and 1 real-time strategy). The first person shooter games typically run at a high frame rate; while real time strategy games don’t scale up the frame rate.

Test Observations and Results

LCD Brightness

A baseline observation that is helpful to understand is based on LCD brightness. Figure 17 shows that a Sony laptop has ~5W (32%) power reduction when using low brightness (0 of 8 bars) as compared to high brightness (8 of 8 bars). The data indicates that reducing screen brightness can help save power. The power saving percentage may change when the system is busy running workloads.

Figure 17: Idle system LCD brightness

Game Resolution Settings

The chosen games support various resolution based on system configuration and game design. Figure 18 shows the effect of power consumption on the Sony laptop with an nVidia 7400 graphics card (Setup #2). The data here represents power scaling, hence lower is better. For example, the blue bar (for game A and B) show baseline/default power consumption data scaled to 1.0. The other bars are scaled accordingly. For game A, with a resolution of 800x600, the power consumption is ~5% less as compared to baseline; when using the 640x480 resolution, the power consumption is ~8% less as compared to baseline. Similar results were observed on the Engineering system (Setup #1.)

Figure 18: Game resolution and average power comparison on a Sony laptop

Game Quality Settings

The chosen games include options to adjust game quality settings, including ‘Textures’, ‘Shaders’, ‘Shadow Effect’, ‘Smoke Effect’, ‘Water Effect’ etc. For example, ‘Shaders’ provide 3 quality settings: high, medium, and low. Shaders usually determine final surface properties of 3D object. It can include details on lighting effect, shadows, reflections, etc and how these properties a ffect the object during play. The details on how much pre-processing is needed (finer level details) can be scaled by selecting high, medium, and low quality shaders.

Figure 19 shows the power consumption impact of changing the game quality settings on Setup #1. The blue bar represents baseline/default power consumption data which is scaled to 1.0. The baseline quality settings change with system configuration and graphics card. The 2nd bar represents measurements taken with all quality settings set to low. As indicated in the graph, setting quality options to low saves power across all 3 games. Similar results were obtained using Setup #2.

Figure 19: Video options and average power consumption on an engineering system

Game Frame Rate Settings

Another option available in the chosen games is an adjustable frame rate. Reducing the frame rate should reduce the necessary processing and thus save on power. Figure 20 shows measurements taken on Setup #2 with varying frame rates. Game A runs at 20 frames per seconds as baseline (blue bar), the maroon bar represents 15 FPS and beige bar presents running at 10FPS. Similarly for game B, the baseline is 60 FPS (blue bar), maroon bar represents 30FPS and beige bar represents running at 20 FPS. As indicated in Figure 20, the platform power consumption is reduced when capping the frame rate at lower values. Similar results were observed on the Engineering system (Setup #1.)

Figure 20: Frame rate and average power consumption on a Sony laptop

Recommendations

It’s clear from the experiments that various game and platform settings can influence the amount of power consumed and when set appropriately, can extend battery time. When running on battery power, the following actions can extend play time:

• Reduce display brightness
• Reduce game resolution to lower value
• Reduce the frame rate by capping it to lower value
• Reduce game quality options to lower values

In addition, if WiFi and/or Bluetooth radios are not needed, power can be saved by turning them off. While these recommendations apply to end users, the game options must first be implemented by the game developers.  This could be as simple as providing a single choice – “Optimized for Performance” or “Optimized for Power.”

Games can be even more user-friendly when they know when to offer choices to the gamer – such as when the user switches from AC to Battery. See the tools section for information on context awareness tools that help developers create applications that can understand the current system state and alter behavior in response to system events.

[1] It is interesting to note that this is not always true. Due to the quadratic relationship between processor states and voltage, it can be demonstrated that a process running for a longer time at a lower P-state may actually use less total energy than running the same process at a high P-state for less time. This is an area of future research.

[2] More detailed coverage of this topic can be found at: http://software.intel.com/en-us/articles/energy-efficient-software

[3] NetDAQ* Networked Data Acquisition Unit

[4] GV3is a Microsoft hotfix (KB896256) to change the kernel power manager to track CPU utilization across the entire package instead of individual cores. It resolves an issue the power manager had with incorrectly calculate the optimal target performance state for the processor when one core was much less busy than the others. The  performance state was set too low and performance suffered in adaptive mode.

[5] More detail on this study can be obtained from: DVD Playback Power Consumption Analysis

[6] For details on MobileMark 2005, see: http://www.bapco.com/

[7] More details from this analysis are available at: Power Analysis of Disk I/O Methodologies

[8] More detailed coverage of this topic can be found at: Data Transfer over Wireless LAN Power Consumption Analysis

[9] For complete details of this study, please see: Enabling Games for Power

[10] For details on Extech Power Analyzers, see: http://www.extech.com/instrument/products/310_399/380803Power.html

[11] See: http://LinuxPowerTOP.org

	Prev	1 2 3  4	Next
Page 3 of 4

For more complete information about compiler optimizations, see our Optimization Notice.

Comments (0)

Trackbacks (0)

Leave a comment