Energy-aware runtime scheduling for embedded-multiprocessor SOCs

Abstract

WITH THE RAPID EVOLUTION of submicronprocess technology, manufacturers are integrating increasing numbers of components on one chip. A heterogeneous system on a chip (SOC) such as that shown in Figure 1 might include one or more programmable components-generalpurpose processor cores, digital signal processor cores, or application-specific intellectual property cores-as well as an analog front end, on-chip memory, I/O devices, and other application-specific ICs. Unfortunately, design technologies have fallen behind advances in processing technology, especially in the context of complex (possibly data-dominated) and very dynamic (partly nondeterministic) applications. SOC designers need a consistent system design technology that can cope with such characteristics and with evershortening time-to-market requirements. This technology should efficiently map these dynamic applications to the target realization while meeting all real-time and other constraints.Contemporary software (and hardware) design practices for this target class can be described only as ad hoc. The design trajecto-

Systematic Design Principles for Cost-Effective Hard Constraint Management in Dynamic Nonlinear Systems

“…It is widely being accepted that cost-effective handling of dynamism requires continuous runtime system adaptation with the help of "knobs" at many abstraction levels and appropriate monitors, leading to the so-called self-adaptive systems (Adve et al, 2002;Cornea et al, 2003) (see Figure 4). This is evident from the growing amount of recent research in the context of low-power systems, on scalable algorithms (Li et al, 2008;Pollin et al, 2007), dynamic power management (Benini et al, 2000), dynamic voltage scaling (Chen et al, 2007;Venkatachalam et al, 2005), MPSoC mapping (Yang et al, 2001;Zhe et al, 2007), etc. Knobs enable run-time Pareto-optimal (see Figure 5) trade-offs between system metrics of interest (such as performance, power consumption, quality, etc.)…”

Atomic Scheduling of Appliance Energy Consumption in Residential Smart Grids

“…Considering that most of the operations subject to DSM are either atomic as such (e.g., laundry cleaning by a clothes washer) or consist of atomic suboperations (e.g., house heating by a heater operating in the morning and in the evening) [11], we provide a new formulation of the optimization problem for atomic scheduling of appliance energy consumption and efficient solution techniques for the resulting problems in this paper. Atomic scheduling has been mostly discussed in the context of concurrent task scheduling on a multiprocessor/core system on a chip (SOC) [18] or transaction processing [19]. To the best of our knowledge, our work is the first attempt to formulate atomic scheduling of appliance energy consumption in the autonomous DSM for residential smart grid, where both the non-interruptibility of appliances' operations and the non-throttleability of the energy consumption patterns are taken into account and guaranteed.…”