Towards better CPU power management on multicore smartphones
Yifan Zhang, Xudong Wang, Xuanzhe Liu, Yunxin Liu, Łi Zhuang, Feng Zhao
Abstract
Yifan Zhang, Xudong Wang, Xuanzhe Liu, Yunxin Liu, Łi Zhuang, Feng Zhao
Abstract
Although multicore smartphones have become increasingly mainstream, it is unclear whether and how smartphone applications can utilize multicore CPUs to improve performance. In this paper we study the performance of mobile applications using multicore CPUs, in terms of power and computation cost. Using Web browsing as an example, our preliminary measurement results show that even large applications like Web browsers with multi-threading acceleration cannot fully utilize the multicore CPUs. Furthermore, we find that the existing CPU power models on smartphones are ill-suited for modern multicore CPUs. We develop a new CPU power model with a high accuracy, 95.6% on average. Our work helps to better understand the performance of multicore smartphones and paves the way towards better CPU power management on multicore smartphones.
OpenAlex reports 39 citations for this work. Citation counts describe recorded attention and do not establish research quality.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
Although multicore smartphones have become increasingly mainstream, it is unclear whether and how smartphone applications can utilize multicore CPUs to improve performance. In this paper we study the performance of mobile applications using multicore CPUs, in terms of power and computation cost. Using Web browsing as an example, our preliminary measurement results show that even large applications like Web browsers with multi-threading acceleration cannot fully utilize the multicore CPUs. Furthermore, we find that the existing CPU power models on smartphones are ill-suited for modern multicore CPUs. We develop a new CPU power model with a high accuracy, 95.6% on average. Our work helps to better understand the performance of multicore smartphones and paves the way towards better CPU power management on multicore smartphones.
Key concepts: Multi-core processor, Computer science, Central processing unit, Power management, Embedded system, Mobile processor, Power (physics), Mobile device