Scholarlink

In this chapter, we will present the concept of system and adaptive scenarios, which are the predecessors of dynamic scenarios. Also, we will deploy a PID scheme on a fine workload granularity, applying Dynamic Voltage and Frequency Scaling (DVFS) actuations per basic workload chunks named as Thread Nodes (TNs); we believe that a TN-level mitigation of performance variation permits an accurate and cost efficient mitigation of performance variability. The structure of this chapter is the following: In Sect. 4.1 we will first define the related terminology and discuss the foundations of the system scenario concept that groups TNs with similar cost perspectives. In Sect. 4.2, we will apply a TN analysis and a system scenario clustering on our spectrum sensing application while presenting related simulation results. The PID controller working in parallel with an adaptive system scenario mechanism, which allows run-time re-clustering decisions of the scenario hierarchy, will be studied later in Sect. 4.3; simulation results of Sect. 4.4 will show the benefit of the latter approach. Finally, Sect. 4.5 will conclude the chapter.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Mitigating Performance Variability Using System and Adaptive Scenarios

  • 章节

  • 发表日期：

    2025年

查看全文
  • Ji-Yung Lin,
  • Michalis Noltsis,
  • Dimitrios Soudris,
  • Francky Catthoor

摘要

In this chapter, we will present the concept of system and adaptive scenarios, which are the predecessors of dynamic scenarios. Also, we will deploy a PID scheme on a fine workload granularity, applying Dynamic Voltage and Frequency Scaling (DVFS) actuations per basic workload chunks named as Thread Nodes (TNs); we believe that a TN-level mitigation of performance variation permits an accurate and cost efficient mitigation of performance variability. The structure of this chapter is the following: In Sect. 4.1 we will first define the related terminology and discuss the foundations of the system scenario concept that groups TNs with similar cost perspectives. In Sect. 4.2, we will apply a TN analysis and a system scenario clustering on our spectrum sensing application while presenting related simulation results. The PID controller working in parallel with an adaptive system scenario mechanism, which allows run-time re-clustering decisions of the scenario hierarchy, will be studied later in Sect. 4.3; simulation results of Sect. 4.4 will show the benefit of the latter approach. Finally, Sect. 4.5 will conclude the chapter.