Buch, Englisch, 89 Seiten, Paperback, Format (B × H): 187 mm x 235 mm
Buch, Englisch, 89 Seiten, Paperback, Format (B × H): 187 mm x 235 mm
Reihe: Synthesis Lectures on Mobile and Pervasive Computing
ISBN: 978-1-59829-128-5
Verlag: MORGAN & CLAYPOOL
The lecture begins by providing the necessary foundations, including basic energy terminology and widely accepted metrics, system models of how power is consumed by a device, and measurement methods and tools available for experimental evaluation. For components that offer low power modes, management policies are considered that address the questions of when to power down to a lower power state and when to power back up to a higher power state. These policies rely on detecting periods when the device is idle as well as techniques for modifying the access patterns of a workload to increase opportunities for power state transitions. For processors with frequency and voltage scaling capabilities, dynamic scheduling policies are developed that determine points during execution when those settings can be changed without harming quality of service constraints. The interactions and tradeoffs among the power management policies of multiple devices are discussed. We explore how the effective power management on one component of a system may have either a positive or negative impact on overall energy consumption or on the design of policies for another component. The important role that application-level involvement may play in energy management is described, with several examples of cross-layer cooperation. Application program interfaces (APIs) that provide information flow across the application-OS boundary are valuable tools in encouraging development of energy-aware applications. Finally, we summarize the key lessons of this lecture and discuss future directions in managing energy demand.
Autoren/Hrsg.
Weitere Infos & Material
- Introduction
- System Energy Models and Metrics
- Management of Device Power States
- Dynamic Voltage Scheduling (DVS)
- Multiple Devices—Interactions and Tradeoffs
- Energy-Aware Application Code
- Challenges and Opportunities