Energy Optimization of Multiprocessor Systems on Chip by Voltage Selection

Abstract

Abstract-Dynamic voltage selection and adaptive body biasing have been shown to reduce dynamic and leakage power consumption effectively. In this paper, we optimally solve the combined supply voltage and body bias selection problem for multi-processor systems with imposed time constraints, explicitly taking into account the transition overheads implied by changing voltage levels. Both energy and time overheads are considered. The voltage selection technique achieves energy efficiency by simultaneously scaling the supply and body bias voltages in the case of processors and buses with repeaters, while energy efficiency on fat wires is achieved through dynamic voltage swing scaling. We investigate the continuous voltage selection as well as its discrete counterpart, and we prove strong NP-hardness in the discrete case. Furthermore, the continuous voltage selection problem is solved using nonlinear programming with polynomial time complexity, while for the discrete problem we use mixed integer linear programming and a polynomial time heuristic. We propose an approach that combines voltage selection and processor shutdown in order to optimize the total energy.

Energy-Efficient Multi- Core Scheduling for Real-Time Video Processing

“…The single-core scheduling [3][4][5][7][8][9][10] executes the real-time task on a single core and turns off the power of the other cores. The all-cores scheduling [13] executes the task on all available cores.…”

“…Numerous previous studies [3][4][5][7][8][9][10] have investigated the energy minimization of real-time tasks on multiple processing elements (PLs). However, they missed the rich energy-saving capability of parallel execution exploiting o verabundant PLs.…”

On the energy optimization for precedence constrained applications using local search algorithms

“…The problem is further complicated because slack allocation must consider variations in workload and energy consumption among different tasks. Moreover, the discrete voltage scaling problem is strongly NP-hard [3], the proof is a reduction to the discrete time-cost trade-off problem by restricting the dynamic voltage scaling problem to contain only tasks that requires an execution of one clock cycle.…”

Minimum-Energy Semi-Static Scheduling of a Periodic Real-Time Task on DVFS-Enabled Multi-Core Processors

“…In accordance with the task's input information obtained at runtime, one of the minimum-energy solutions found at off-line time is applied to each real-time task. Moreover, the proposed scheme is designed to operate over discretely available frequencies with their arbitrary energy consumptions and considers the energy overhead for activating/inactivating cores, whereas the previous methods are designed over infinitely continuous frequencies with an enforced energy consumption formula [1], [2], [4]- [6], [8] and do not consider the energy overhead for activating/inactivating cores [5], [6], [8], [9].…”