Using Associative CPU Utilization to Construct Multi-core Power Estimation Model
2014
Hochschulschrift
Zugriff:
102
Smartphones are wide spread and the number of smartphone users and applications are continuously increasing. One of the most nerve-wracking concerns to the users is full discharge of battery. Since there are more multi-core smartphones, it becomes important to measure the power consumption of multi-core smartphones. In order to improve the performance, the multi-core smartphones use multiple CPUs to share the workload with parallel processing. However, the CPU utilization will influence each other in these devices. Previous studies assume the CPU utilization in smartphones as an independent parameter, causing their power consumption is not accurately estimated. In this thesis, we propose a Multi-core Power Estimation Model (MPEM) according to a Utilization Based approach (UB). MPEM first measures the power consumption and CPU utilization, and then uses the regression analysis to build an effective power consumption model. To more precisely consider the power consumption factors on CPU, MPEM uses the associate CPU utilization. MPEM has two power formulas, Dependent CPU Utilization (DCU) and Load-based CPU Utilization (LCU). Experiment results show that DCU has decrease 5%~10% error rate, compared with the previous work. Also LCU has only 10% error rate, that is, it has a lower error rate about 5%~10% than DCU.
Titel: |
Using Associative CPU Utilization to Construct Multi-core Power Estimation Model
|
---|---|
Autor/in / Beteiligte Person: | Huang, Yu-shan ; 黃郁珊 |
Link: | |
Veröffentlichung: | 2014 |
Medientyp: | Hochschulschrift |
Sonstiges: |
|