Multimedia power management on a platter: from audio to video & games
MM '08 Proceedings of the 16th ACM international conference on Multimedia
Aggressive dynamic voltage scaling for energy-aware video playback based on decoding time estimation
EMSOFT '09 Proceedings of the seventh ACM international conference on Embedded software
ARIVU: power-aware middleware for multiplayer mobile games
Proceedings of the 9th Annual Workshop on Network and Systems Support for Games
Adaptive display power management for mobile games
MobiSys '11 Proceedings of the 9th international conference on Mobile systems, applications, and services
Power gating strategies on GPUs
ACM Transactions on Architecture and Code Optimization (TACO)
A QoS-aware memory controller for dynamically balancing GPU and CPU bandwidth use in an MPSoC
Proceedings of the 49th Annual Design Automation Conference
Energy-aware adaptations in mobile 3d graphics
Proceedings of the 20th ACM international conference on Multimedia
Energy efficient multi-player smartphone gaming using 3D spatial subdivisioning and pvs techniques
Proceedings of the 3rd ACM international workshop on Interactive multimedia on mobile & portable devices
A resource-driven DVFS scheme for smart handheld devices
ACM Transactions on Embedded Computing Systems (TECS)
Hi-index | 0.00 |
Interactive 3D games are now widely available on a variety of mobile devices for which battery-life is a major concern. Many of these devices support voltage/frequency-scalable processors and dynamic voltage scaling (DVS) has emerged as a powerful technique for energy management in such devices. Although DVS algorithms have been very successfully applied to video encoding/decoding applications, their use in interactive computer games has not been sufficiently explored so far. In this paper we propose a novel DVS scheme that is specifically directed towards interactive 3D game applications running on battery-operated portable devices. The key to this DVS scheme lies in an accurate prediction of the rendering workload of a current game scene. We have applied this scheme to First Person Shooter games (e.g. Quake II) and obtained significant power savings while maintaining high frame rates. Based on the observation that there exist two types of workload variations in such games, we compute the voltage/frequency setting for any game scene using a hybrid combination of two different techniques: (i) adjusting the workload prediction using a control-theoretical feedback mechanism, and (ii) analyzing the graphical objects in the current game scene by parsing the corresponding frame. Our scheme is significantly different from those commonly applied to video decoding applications (where only technique (i) is used) and has shown very encouraging results when evaluated with different setups (e.g. laptop running Windows, PDA running Windows Mobile and a configurable simulation platform).