Dynamic voltage frequency scaling (DVFS), as explained in this book, allows processors to dynamically adjust supply voltage or clock frequency to operate on different energy levels. This is considered an effective way to save energy. Scheduling is a well-known nondeterministic polynomial-time (NP)-hard problem, and this short book surveys current research on scheduling heuristics for energy-aware multiprocessor platforms. It is first and foremost a scheduling survey, where different levels of power consumption are being considered as another constraint on the optimization process. The problem, then, is either defined as minimizing energy consumption to execute the tasks at hand by some deadline or, conversely, to maximize throughput within a certain power consumption threshold.
The book has six chapters. After a brief introduction, chapter 2 formalizes the main concepts used throughout the book, particularly tasks and platform models. It also defines all the possible scheduling events considered by the model, such as slack reclamation, task migration, and so on.
Chapter 3 presents a collection of proposed scheduling heuristics for homogeneous DVFS multiprocessor platforms, classified by task type. Chapter 4 covers heuristics for heterogeneous platforms. Chapters 3 and 4 do not provide any analysis or comparisons of the many heuristics described, which would have produced a more complete approach.
Chapter 5 surveys related work and comments on other surveys on energy-aware scheduling. Chapter 6 concludes the book with a discussion on future work.
Unfortunately, the book did not receive a thorough editorial review. There is no cohesion of style and the nomenclature varies from section to section. The standards shown in Table 2.1 (“Notations Used in This Book”) are not always followed. For example, processors are sometimes referred to as μ, and sometimes as π, and tasks, although mostly τ, are sometimes tk. The abstract to chapter 4 is an out-of-control list of references spelled out in full form, even though they are referred to appropriately later in the chapter.