Warp-Aware Adaptive Energy Efficiency Calibration for Multi-GPU Systems
Zhuowei Wanga b, Xiaoyu Song Bc, Lianglun Chenga, Hai Wand, Wuqing Zhaoe, Tao Wangf
- aSchool of Computers, Guangdong University of Technology, Guangzhou, China
- bSchool of Computer Science, Wuhan Donghu University, Wuhan, China
- cDepartment of Electrical and Computer Engineering, Portland State University, Portland, OR, USA
- dSchool of Software, Tsinghua University, Beijing, China
- eDigital Grid Research Institute, China Southern Power Grid, Guangzhou, China
- fSchool of Automation, Guangdong University of Technology, Guangzhou, China
Papers open inside SciRah from the original publisher or repository; SciRah never hosts them.
Abstract
Massive GPU acceleration processors have been used in high-performance computing systems. The Dennard scaling has led to power and thermal constraints limiting the performance of such systems. The demand for both increased performance and energy efficiency is highly desired. This article presents a multilayer low-power optimization method for warps and tasks parallelisms. We present a dynamic frequency regulation scheme for performance parameters in terms of load balance and load imbalance. The method monitors the energy parameters in runtime and adjusts adaptively the voltage level to ensure performance efficiency with energy reduction. The experimental results show that the multilayer low-power optimization with dynamic frequency regulation can achieve 40% energy consumption reduction with only 1.6% performance degradation, thus reducing 59% maximum energy consumption. It can further save about 30% energy consumption in comparison with the single-layer energy optimization.