作者: Jie Ren , Xiaoming Wang , Jianbin Fang , Yansong Feng , Dongxiao Zhu
关键词:
摘要: We present Proteus, a novel network-aware approach for optimizing web browsing on heterogeneous multi-core mobile systems. It employs machine learning techniques to predict which of the cores use render given webpage and operating frequencies processors. achieves this by first offline set predictive models range typical networking environments. A learnt model is then chosen at runtime optimal processor configuration, based content, network status optimization goal. evaluate Proteus implementing it into open-source Chromium browser testing two representative ARM big.LITTLE platforms. apply top 1,000 popular websites across seven over 80% best available performance. obtains, average, 17% (up 63%), 31% 88%), 30% 91%) improvement respectively load time, energy consumption delay product, when compared state-of-the-art approaches.