Profil Energi Pada Processor Perangkat Edge Computing Berbasis DVFS dalam Menjalankan Program Paralel

Abstract

This study aims to estimate the processor energy profile of an edge computing device during parallel program execution across various operating frequencies, utilizing Dynamic Voltage and Frequency Scaling (DVFS).
The SparseLU and Strassen algorithms from the Barcelona OpenMP Tasks Suite (BOTS) were executed on an Intel NUC DN2820FYK PC using the Intel P-State driver. Both programs were run at different Intel NUC operating frequencies determined by DVFS principles. Processor energy consumption was estimated to establish the relationship between energy usage and operating frequency.
This research successfully generated processor energy usage profiles for SparseLU and Strassen across the various operating frequencies defined by DVFS. Furthermore, energy consumption efficiency was achieved by increasing the CPU frequency—resulting in faster execution times—even though the power consumed was higher.