Dynamic Voltage And Frequency Scaling Algorithms Research Articles

Dynamic frequency scheduling for CubeSat's on-board and data handling subsystem

<p>CubeSat is a small-sized satellite that provides a cheaper option for the manufacturer to have a fully operational satellite. Due to its size, CubeSat can only generate limited power, and this will restrict its functionality. This research aims to improve CubeSat’s power consumption by implementing the dynamic voltage and frequency scaling (DVFS) technique to on-board and data handling subsystem (OBDH). DVFS will find the best operating frequency to execute all of OBDH’s task. This paper explains how we determined the task set, representing all routine tasks performed by OBDH during normal operation mode. We have simulated the task set using two DVFS algorithms, static earliest deadline first (EDF) and cycle conserving edf (CC EDF). The result shows that both scheduling algorithms give a similar result to our task set. However, when the scheduler is configured as non-preemptive, the simulator failed to schedule the critical task. It means that the system fails to work as intended. Therefore, we conclude that we need to implement mixed-criticality scheduling to prevent critical tasks from being aborted by the system.</p>

A Double Deep Q-Learning Model for Energy-Efficient Edge Scheduling

Reducing energy consumption is a vital and challenging problem for the edge computing devices since they are always energy-limited. To tackle this problem, a deep Q-learning model with multiple DVFS (dynamic voltage and frequency scaling) algorithms was proposed for energy-efficient scheduling (DQL-EES). However, DQL-EES is highly unstable when using a single stacked auto-encoder to approximate the Q-function. Additionally, it cannot distinguish the continuous system states well since it depends on a Q-table to generate the target values for training parameters. In this paper, a double deep Q-learning model is proposed for energy-efficient edge scheduling (DDQ-EES). Specially, the proposed double deep Q-learning model includes a generated network for producing the Q-value for each DVFS algorithm and a target network for producing the target Q-values to train the parameters. Furthermore, the rectified linear units (ReLU) function is used as the activation function in the double deep Q-learning model, instead of the Sigmoid function in QDL-EES, to avoid gradient vanishing. Finally, a learning algorithm based on experience replay is developed to train the parameters of the proposed model. The proposed model is compared with DQL-EES on EdgeCloudSim in terms of energy saving and training time. Results indicate that our proposed model can save average $2\%\hbox{-}2.4\%$2%-2.4% energy and achieve a higher training efficiency than QQL-EES, proving its potential for energy-efficient edge scheduling.

DVFS Algorithms of GPU and Memory for Mobile GPGPU Applications: A case study

Although both OpenCL and RenderScript have allowed the General-Purpose Graphics Processing Unit (GPGPU) to be used even for mobile GPUs, it is still difficult for mobile applications to use the GPGPU for several reasons. One of the reasons is that mobile devices place restrictions on GPU performance through power-saving technologies such as Dynamic Voltage and Frequency Scaling (DVFS). DVFS tries to control the balance between performance and energy consumption based on the application’s requirements. This technology has been successful in many cases and is widely used; however, it significantly decreases the performance of GPGPU applications. In this paper, we propose novel DVFS algorithms for GPU and memory when the GPGPU applications run. The suggested algorithms decreased the energy consumption by more than 0.7 times without any algorithm changes, and improved the energy efficiency (performance per watt) by more than 3.42 times in comparison with the conventional interval-based algorithm.

Impact of Cache Voltage Scaling on Energy-Time Pareto Frontier in Multicore Systems

High performance computing centers need to keep up with the growing workload of varying computational characteristics. Due to their high computation rates, these computing systems consume vast amounts of energy with increasing electricity costs. As an approach to balancing computation demand with energy consumption, state-of-the-art dynamic voltage and frequency scaling (DVFS) methodologies are used for improving the energy efficiency of computing systems. However, these studies often do not explore the extent to which their solutions are close to theoretically optimal limits. This work formulates optimal boundaries for energy-time performance with a Linear Programming (LP) approach. The formulation utilizes per-core energy consumptions and execution times obtained during the profiling phase to optimize the voltage and frequency (V/F) level assignments at runtime. For each of the four benchmarks considered in this work, the optimized V/F assignments are used to bound Pareto frontiers, which trade off energy consumption and execution time. In particular, this work studies the impact of scaling the voltage and frequency of the cache subsystem in a multicore system on establishing the energy-time Pareto frontier. An unexpected result of our study is that when the frequencies of caches are not scaled with that of the cores (i.e., fixed at 2.0GHz), the proposed LP-based technique improves the overall Energy-Delay-Product (EDP) as much as 35% compared to the traditional no-DVFS Pareto frontier. Furthermore, this work compares the performance of three heuristic-based energy-efficient DVFS algorithms to demonstrate the differences between heuristics performances and the LP-based optimal Pareto frontier.

Energy-efficient global illumination algorithms for mobile devices using dynamic voltage and frequency scaling

With increases of display resolution and graphical quality demands, global illumination techniques are sought after to meet such demands. However, global illumination algorithms have a long processing time on mobile devices, such as smartphones or tablet PCs. Besides the performance issue, the algorithms consume a large amount of battery-powered energy. The performance and energy consumption have a trade-off relationship. The dynamic voltage and frequency scaling (DVFS) algorithms control the balance between the performance and the energy consumption by adjusting the GPU and CPU’s frequencies. To improve the performance of the global illumination algorithm and reduce the high energy consumption of the current DVFS algorithm, we suggest new algorithms with new metrics for estimating the amount of workload for GPUs as well as their memory workload. Using our new DVFS algorithms, we increase the energy efficiency, the performance per watt, of a global illumination algorithm by 33.59% in comparison to a conventional general-purpose DVFS algorithm, without modifying the underlying algorithm core.