Opportunities and Challenges of Application-Power Control in the Age of Dark Silicon
スポンサーリンク
概要
- 論文の詳細を見る
In the age of dark silicon on-chip power control is a necessity. Upcoming and state of the art embeddedand cloud computer system-on-chips (SoCs) already provide interfaces for fine grained power control. Sometimes both: core- and interconnect-voltage and frequency can be scaled for example. To further reduce power consumption SoCs often have specialized accelerators. Due to the rising specialization of hard- and software general purpose operating systems require changes to exploit the power saving opportunities provided by the hardware. However, they lack detailed hardware- and application-level-information. Application-level power control in turn is still very uncommon and difficult to realize. Now a days vendors of mobile devices are forced to tweak and patch system-level software to enhance the power efficiency of each individual product. This manual process is time consuming and must be reiterated for each new product. In this paper we explore the opportunities and challenges of automatic application-level power control using compilers.
- 2012-12-06
著者
-
Keiji Kimura
Waseda University
-
Dominic Hillenbrand
Waseda University
-
Yuuki Furuyama
Waseda University
-
Akihiro Hayashi
Waseda University
-
Hiroki Mikami
Waseda University
-
Hironori Kasahara
Waseda University
-
Keiji Kimura
Department of Computer Science and Engineering, Waseda University
関連論文
- Opportunities and Challenges of Application-Power Control in the Age of Dark Silicon
- Automatic parallelization with OSCAR API Analyzer: a cross-platform performance evaluation
- Opportunities and Challenges of Application-Power Control in the Age of Dark Silicon
- Automatic parallelization with OSCAR API Analyzer: a cross-platform performance evaluation
- Enhancing the Performance of a Multiplayer Game by Using a Parallelizing Compiler
- Enhancing the Performance of a Multiplayer Game by Using a Parallelizing Compiler
- Android Demonstration System of Automatic Parallelization and Power Optimization by OSCAR Compiler