Name
Abstract | ||
|---|---|---|
In recent years, many studies on energy-efficient optimization with performance constraint have focused on “power-performance” assignment in multi-core processors. To find the solution of “power-performance” assignment, progressive search for optimal configuration of core number and frequency can be implemented in two dimensions (core number and frequency) according to power-performance models. However, the existing methods of searching for optimal energy-efficient configuration in the space of core number and frequency have the faults of slow convergence speed, great overhead and poor scalability. In this paper, a search method based on feasible direction method is developed to quickly reduce search space in the two dimensions of core number and frequency and converge to the minimum point of energy consumption in the iterative process. Simultaneously, the practical energy and performance of each feasible configuration can revise model computation. The experimental results show that compared with Hill-climbing Heuristic, the best one of the existing search methods for energy-efficient optimization, in the number of execution, execution time and energy overhead, our framework makes an average reduction by 38.6%, 43.9% and 46.7% respectively, 47.6%, 50.2% and 49.3% when doubling cores of a multi-core processor, and 44.7%, 49.1%, 53.2% when doubling the frequency levels. |
Year | DOI | Venue |
|---|---|---|
2016 | 10.1109/IGCC.2016.7892616 | 2016 Seventh International Green and Sustainable Computing Conference (IGSC) |
Keywords | Field | DocType |
energy-efficient optimization,multi-cores,power-performance,feasible direction method,Hill-climbing Heuristic | Convergence (routing),Heuristic,Iterative and incremental development,Efficient energy use,Computer science,Parallel computing,Linear programming,Energy consumption,Multi-core processor,Scalability | Conference |
ISBN | Citations | PageRank |
978-1-5090-5118-2 | 0 | 0.34 |
References | Authors | |
14 | 8 | |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Yatao Zhu | 1 | 0 | 0.68 |
| Xiaochun Ye | 2 | 125 | 28.41 |
| Da Wang | 3 | 44 | 8.79 |
| Wenming Li | 4 | 20 | 5.86 |
| Yang Zhang | 5 | 0 | 0.34 |
| FAN Dong-Rui | 6 | 222 | 38.18 |
| Zhimin Zhang | 7 | 54 | 11.10 |
| Zhimin Tang | 8 | 234 | 22.55 |