A unified practical approach to stochastic DVS scheduling

Abstract

This paper deals with energy-aware real-time system scheduling using dynamic voltage scaling (DVS) for energy-constrained embedded systems that execute variable and unpredictable workloads. The goal is to design DVS schemes to minimize the expected energy consumption of the whole system while meeting the deadlines of the tasks. Researchers have attempted to take advantage of stochastic information about workloads to achieve better energy savings, and accordingly, various stochastic DVS schemes have been proposed. However, the existing stochastic DVS schemes are based on much simplified power models that assume unrestricted continuous frequency, well-defined power/frequency relation, and no speed change overhead. When these schemes are used in practice, they need to be patched in order to comply with realistic power models. Experiments show that some of such DVS schemes perform even worse than certain non-stochastic DVS schemes. Furthermore, even for stochastic schemes that were shown experimentally to outperform non-stochastic schemes, it is not clear how well they perform compared to the optimal solution, which is yet to be found. In this work, we provide a unified practical approach for obtaining optimal (or provably close to optimal) stochastic inter-task, intra-task, and hybrid DVS schemes under realistic power models in which the processor only provides a set of discrete speeds, no assumption is made on power/frequency relation, and speed change overhead is considered. We also evaluate the existing DVS schemes by comparing them with our DVS schemes.

Time Series Characterization of Gaming Workload for Runtime Power Management

“…A considerable amount of work on runtime processor-energy management has viewed this problem from a scheduling perspective [5]- [7], [41], [43], i.e., how to select the appropriate clock frequency (which might include shutting off the processor) and voltage at each time step in order to maximize energy savings, taking into account issues like switching overhead and application quality. There has also been work on scheduling and energy management of multiple components like (multi-core) processors and communication links [13], [27].…”

Energy efficient multiprocessor task scheduling under input-dependent variation

“…The work in [7][9] [12] [14][28] models the execution time of a task as a random variable and minimizes expected energy consumption on a single processor system. A heuristic is provided in [7] for obtaining a low-energy schedule.…”

“…In [9] [12], exact solutions are provided using convex optimization techniques; however, many of their assumptions, such as the ability to change the voltage to any arbitrary value at any point during the execution of a task, are not valid for practical systems. Many of these issues are addressed in [14] for uniprocessor systems. In [28], a mathematical formulation is presented to optimize the expected energy consumption (both dynamic and leakage) by using DVFS and Adaptive Body Biasing (ABB).…”

Efficient Algorithms for Extracting Pareto-optimal Hardware Configurations in DEPS Framework

“…Azevedo et al first proposed a method using a checkpoint function for voltage scheduling [5]. Xu et al provided a unified practical approach for obtaining optimal stochastic inter-task, intratask, and hybrid DVFS schemes [6]. Bambha et al proposed a hybrid global/local configuration search for DVFS in embedded multiprocessor systems [7].…”