A measurement study of GPU DVFS on energy conservation

Xinxin Mei, Ling Sing Yung, Kaiyong Zhao, Xiaowen Chu
2013 Proceedings of the Workshop on Power-Aware Computing and Systems - HotPower '13  
Energy conservation on computing systems has become an important research topic in recent years. Dynamic voltage/frequency scaling (DVFS) has been shown as an appealing method for saving energy. Nowadays, GPUs have been widely used to accelerate many high performance computing applications. However, there is a lack of study on the effectiveness of GPU DVFS on energy conservation. This paper presents a thorough measurement study that aims to explore how GPU DVFS affects the system energy
more » ... stem energy consumption. We have conducted experiments on a real GPU platform with 37 benchmark applications. Our results show that GPU DVFS is an effective approach to conserving energy. For example, by scaling down the GPU core voltage and frequency, we have achieved an average of 19.28% energy reduction compared with the default setting, while giving up no more than 4% of performance. For all tested GPU applications, core voltage scaling is significantly effective to reduce system energy consumption. On the other hand, the effect of scaling core frequency and memory frequency depends on the characteristic of GPU applications.
doi:10.1145/2525526.2525852 dblp:conf/sosp/MeiYZC13 fatcat:6ayk3y7xhzfhjcsb4hgdfu4muq