Loading...
Please wait, while we are loading the content...
Similar Documents
Energy-Aware Resource Adaptation in the Tessellation OS
| Content Provider | Semantic Scholar |
|---|---|
| Author | Chou, David Eads, Gage |
| Copyright Year | 2012 |
| Abstract | Modern devices must juggle the competing requirements of parallel high-throughput, interactive, and multimedia applications while conserving energy where possible. Each requires some combination of CPU, cache, memory, and network bandwidth, and allocating resources among applications is a complex optimization problem. Application requirements are often conflicting, and users wants predictable responsiveness, and long battery life, despite dynamically changing resources. Modern OSes do not adequately tackle this problem; system-wide power goals do not consider the priorities of various applications, and can hurt the user experience. We present a resource adaptation system, developed in the Tessellation OS, that model and schedule resources for energy consumption, in order to automatically find low-energy and high-performance resource configurations for realistic workloads. Our results show that in a two app scenario with differing priorities, our system automatically finds a resource allocation that meets both performance requirements or (when power saving is more important) reduces system power by 7 W and continues to meet the high priority application’s performance target. |
| File Format | PDF HTM / HTML |
| Alternate Webpage(s) | https://people.eecs.berkeley.edu/~kubitron/courses/cs262a-F12/projects/reports/project10_report.pdf |
| Language | English |
| Access Restriction | Open |
| Content Type | Text |
| Resource Type | Article |
