System-Level Power Management And An Overview:Background

Introduction

One of the key challenges of computer system design is the management and conservation of energy. This challenge is evident in a number of ways. The goal may be to extend the battery lifetime of a portable, battery-powered device. The processing power, memory, and network bandwidth of such devices are increasing quickly, resulting in an increase in demand for higher power dissipation, while the battery capacity is improving at a much slower pace. Other goals may be to limit the cooling requirements of a computer system or to reduce the financial cost of operating a large computing facility with a high energy bill. This chapter focuses on techniques which dynamically manage electronic systems in order to minimize its energy consumption. Ideally, the problem of managing the energy consumed by electronic systems should be addressed at all levels of design, ranging from low-power circuits and architectures to application and system software capable of adapting to the available energy source. Many research and industrial efforts are currently underway to develop low-power hardware as well as energy-aware application software in the design of energy-efficient computing systems. Our objective in this chapter is to explore what the system software, vis-à-vis the operating system (OS), can do within its own resource management functions to improve the energy efficiency of the computing system without requiring any specialized, low-power hardware or any explicit assistance from application software and compilers. There are two approaches to consider at the OS-level for attacking most of the specific energy-related goals described above. The first is to develop resource management policies that eliminate waste or overhead and allow energy-efficient use of the devices. The second is to change the system workload so as to reduce the amount of work to be done, often by changing the fidelity of objects accessed, in a manner which will be acceptable to the user of the application. This chapter provides a first introduction to these two approaches with appropriate review of related works.

Background

A system is a collection of components whose combined operation provides a useful service. Typical systems consist of hardware components integrated on single or multiple chips and various software layers. Hardware components are macro-cells that provide information processing, storage, and interfacing. Software components are programs that realize system and application functions. Sometimes, system specifications are required to fit into specific interconnections of selected hardware components (e.g., Pentium processor) with specific system software (e.g., Windows or Linux) called computational platforms.

System design consists of realizing a desired functionality while satisfying some design constraints. Broadly speaking, constraints limit the design space and relate the major design trade-off between quality of service (QoS) versus cost. QoS is closely related to performance, i.e., system throughput and task latency. QoS relates also to the system dependability, i.e., to a class of system metrics such as reliability, availability, and safety that measure the ability of the system to deliver a service correctly, within a given time window and at any time. Design cost relates to design and manufacturing costs (e.g., silicon area and testability) as well as to operation costs (e.g., power consumption and energy consumption per task). In recent years, the design trade-off of performance versus power consumption has received large attention because of (i) the large number of systems that need to provide services with the energy provided by a battery of limited weight and size, (ii) the limitation on high-performance computation because of heat dissipation issues, and (iii) concerns about dependability of systems operating at high temperatures because of power dissipation. Here we focus on energy-managed computer (EMC) systems. These systems are characterized by one or more high-performance processing cores, large on-chip memory cores, various I/O controller cores. The use of these cores will force system designers to treat them as black boxes and abandon the detailed tuning of their performance/energy parameters. On the other hand, various I/O devices are provisioned in the system-level design to maximize the interaction between the user and the system and among different users of the same system.

Dynamic power management (DPM) is a feature of the run-time environment of an EMC system that dynamically reconfigures itself to provide the requested services and performance levels with a minimum number of active components or a minimum activity level on such components. DPM encompasses a set of techniques that achieve energy-efficient computation by selectively turning off (or reducing the performance of) system components when they are idle (or partially unexploited). The fundamental premise for the applicability of DPM is that systems (and their components) experience nonuniform workloads during operation time. Such an assumption is valid for most systems, both when considered in isolation and when internetworked. A second assumption of DPM is that it is possible to predict, with a certain degree of confidence, the fluctuations of workload. In this chapter we present and classify different modeling frameworks and approaches to DPM.