SmartBoost: Lightweight ML-Driven Boosting for Thermally-Constrained Many-Core Processors
TimeTuesday, December 7th1:30pm - 1:52pm PST
Event Type
Research Manuscript
Virtual Programs
Presented In-Person
Cross-Layer Power Analysis and Low-Power Design
DescriptionDynamic voltage and frequency scaling (DVFS)-based boosting is indispensable for optimizing the performance of a thermally-constrained many-core processor. This paper introduces a novel boosting metric that for the first time integrates the application-dependent voltage/frequency (v/f) sensitivities of performance and power, and the core-dependent v/f sensitivity of the temperature. This new boosting metric is derived at run-time via machine-learning-based estimates of the v/f sensitivities of performance and power of unknown applications with diverse and time-varying characteristics. We build a smart lightweight boosting technique based on the new metric. The experimental results demonstrate a significant performance improvement compared to the state-of-the-art.