Power, Energy, and Thermal Management for Clustered Manycores
Santiago Pagani · Repository KITopen (Karlsruhe Institute of Technology) · 2016
Efficient and effective system-level power, energy, and thermal management are very important issues in modern computing systems, for which clustered architectures with multiple voltage islands are an expected compromise between global and per-core DVFS. In this dissertation, we focus on two of the most relevant problems for such architectures, specifically, optimizing performance under power/thermal constraints, and minimizing energy under performance constraints.