Evaluation of DVFS techniques on modern HPC processors and accelerators for energy-aware applications

摘要

Energy efficiency is becoming increasingly important for computing systems,in particular for large scale HPC facilities. In this work we evaluate, from anuser perspective, the use of Dynamic Voltage and Frequency Scaling (DVFS)techniques, assisted by the power and energy monitoring capabilities of modernprocessors in order to tune applications for energy efficiency. We run selectedkernels and a full HPC application on two high-end processors widely used inthe HPC context, namely an NVIDIA K80 GPU and an Intel Haswell CPU. We evaluatethe available trade-offs between energy-to-solution and time-to-solution,attempting a function-by-function frequency tuning. We finally estimate thebenefits obtainable running the full code on a HPC multi-GPU node, with respectto default clock frequency governors. We instrument our code to accuratelymonitor power consumption and execution time without the need of any additionalhardware, and we enable it to change CPUs and GPUs clock frequencies whilerunning. We analyze our results on the different architectures using a simpleenergy-performance model, and derive a number of energy saving strategies whichcan be easily adopted on recent high-end HPC systems for generic applications.