基于Android系统的GPU动态调频方案

Dynamic frequency scaling solution of GPU based on Android system

针对移动终端上GPU的高功耗问题,提出一种基于Android系统的GPU动态调频方案。方案根据各种应用对GPU的性能需求,引入了GPU的频率-性能模型,包括选择工作频率和测量相对性能的方法。动态调频算法通过历史负载计算出预测负载,将其代入频率-性能模型后预测出下一周期GPU的频率。实验结果表明,方案在典型场景下可以快速跟踪GPU负载的变化,预测GPU频率的准确率达到95%以上。

A dynamic frequency scaling solution of GPU based on Android system was proposed in order to resolve the high power problem of GPU in mobile terminals. A GPU frequency-performance model, including the methods of working frequency selecting and relative performance measuring, was introduced according to the GPU performance requirement of various applications. In the DVFS algorithm, the prediction load was calculated by weighted previous loads and substituted into the GPU frequency-performance model to predict the frequency in next period. The experimental results show that the changing of GPU load can be tracked rapidly in typical cases, and the accuracy of GPU frequency prediction is more than 95 %.