In this paper we investigate the problem of passivity-based sliding mode control for fractional-order hyperchaotic systems with unknown parameters. Utilizing the fractional calculus, the fractional-order sliding surfa...
详细信息
The study recommends utilizing image processing techniques and a dual-layer CNN for analyzing datasets related to root rot. The research emphases on detecting and diagnosing root rot over Machine Learning and image pr...
详细信息
Over the last decade both capital costs for the acquisition of modern HPC systems as well as costs to power and cool these systems have increased significantly. In addition to direct operational costs, the investment ...
详细信息
ISBN:
(纸本)9798350383461;9798350383454
Over the last decade both capital costs for the acquisition of modern HPC systems as well as costs to power and cool these systems have increased significantly. In addition to direct operational costs, the investment costs for the infrastructure required to provide the necessary energy and cooling capacity have also risen substantially. In order to shift the ratio between infrastructure costs and system costs in favor of the latter, the High Performance computing Center Stuttgart (HLRS) has decided to size the system Hawk based on energy requirements of typical user workloads. To not overload the data centre's infrastructure the power consumption of the Hawk system has to be constrained. The Hawk system hence is operated as an over-provisioned system where nameplate thermal design power exceeds provisioned power. In this work we explain how the dynamic power management solution called PowerSched which is implemented by Hewlett Packard Enterprise (HPE) at HLRS, is able to balance compute jobs with moderate demands for CPU-power against CPU-power hungry jobs and thus keeps the overall power consumption of the system and its components within the required limits. We elaborate how the solution minimizes power consumption while maximizing the overall performance of all applications executed in the system at any given time. We further demonstrate the effectiveness of the chosen approach and the effects of PowerSched on power consumption at system and component level. Finally, the performance impact at application level is compared against executions with static power capping per CPU and against executions with unconstrained power.
Approximate computing offers a solution since it is a potential technique to improve area, speed, and reduce power in cases when accurate calculation is not expected. In this work, an unsigned approximate multiplier a...
详细信息
Machine Learning is increasingly crucial for predicting application performance, offering a black-box approach that does not require a deep understanding of the application internal workings. This method enables accur...
详细信息
ISBN:
(纸本)9798350389777;9798350389760
Machine Learning is increasingly crucial for predicting application performance, offering a black-box approach that does not require a deep understanding of the application internal workings. This method enables accurate predictions without delving into complex system models. Our study utilized ML to forecast the execution time of an industrial application dealing with risk measures as part of the Solvency II regulations for insurance companies. By conducting a comparative analysis of multiple models, XGBoost was identified as the most effective, achieving a Mean Absolute Percentage Error of 18%. The results demonstrated robust accuracy for intermediate durations, though limitations were observed for shorter and significantly longer times due to data scarcity. Overall, this study highlights the significant potential of ML in improving prediction accuracy for complex industrial applications, offering valuable insights for resource allocation and performance management.
Spatial co-location pattern mining (SCPM) is a sub-field of data mining, which aims to discover the subset of spatial features whose instances are frequently located in proximate areas. SCPM has broad prospects in man...
详细信息
With the increasing penetration of distributed photovoltaic (PV) in low-voltage distribution network (LVDN), the operational risks of LVDN are also increasing. To address the issue of insufficient security and stabili...
详细信息
A growth in data volume, combined with increasing demand for real-time analysis (using the most recent data), has resulted in the emergence of database systems that concurrently support transactions and data analytics...
详细信息
ISBN:
(数字)9781665408837
ISBN:
(纸本)9781665408837
A growth in data volume, combined with increasing demand for real-time analysis (using the most recent data), has resulted in the emergence of database systems that concurrently support transactions and data analytics. These hybrid transactional and analytical processing (HTAP) database systems can support real-time data analysis without the high costs of synchronizing across separate single-purpose databases. Unfortunately, for many applications that perform a high rate of data updates, state-of-the-art HTAP systems incur significant losses in transactional (up to 74.6%) and/or analytical (up to 49.8%) throughput compared to performing only transactional or only analytical queries in isolation, due to (1) data movement between the CPU and memory, (2) data update propagation from transactional to analytical workloads, and (3) the cost to maintain a consistent view of data across the system. We propose Polynesia, a hardware-software co-designed system for in-memory HTAP databases that avoids the large throughput losses of traditional HTAP systems. Polynesia (1) divides the HTAP system into transactional and analytical processing islands, (2) implements new custom hardware that unlocks software optimizations to reduce the costs of update propagation and consistency, and (3) exploits processing-in-memory for the analytical islands to alleviate data movement overheads. Our evaluation shows that Polynesia outperforms three state-of-the-art HTAP systems, with average transactional/analytical throughput improvements of 1.7x/3.7x, and reduces energy consumption by 48% over the prior lowest-energy HTAP system.
Intelligent Transportation System (ITS) is embarking on a new development road, in which Digital Twins (DT) provides crucial support for improving the smartness standard of vehicles and realizing a high-level of traff...
详细信息
With the expansion of connected devices used for industrial purposes, the Industrial Internet of Things (IIoT) has emerged as a specific branch of the Internet of Things (IoT) for Industry 4.0. Its applications in ind...
详细信息
ISBN:
(纸本)9781538674628
With the expansion of connected devices used for industrial purposes, the Industrial Internet of Things (IIoT) has emerged as a specific branch of the Internet of Things (IoT) for Industry 4.0. Its applications in industrial domains include monitoring, smart manufacturing, and virtual and augmented reality. However, the huge amount of data generated by IIoT devices with limited computing resources makes it challenging to guarantee the different required quality of service (QoS) of the applications while minimizing the computation cost in terms of energy consumption. especially in terms of latency. In this paper, we opt for the offloading of dependent computation-intensive tasks to edge and cloud servers with more powerful computation capacities. Our proposed model aims to minimize the energy consumption of each IIoT device while respecting the maximal tolerant deadline of task completion. We propose to cluster the IIoT devices that have dependent tasks together to better handle this dependency. Moreover, we propose a distributed cooperative game that allows each device to decide whether it is beneficial for it and for its cluster, in terms of task completion and energy consumption, to offload its task or execute it locally. We prove that the Nash Equilibrium exists by proving that our game is a weighted potential game. Finally, we propose a practical distributed offloading algorithm to implement the cooperative game. The performance evaluation results show that the proposal optimizes energy consumption whilst increasing the number of tasks completed on time.
暂无评论