검색 상세

Energy and Latency Co-aware Management Techniques for GPU-based Stream Processing Systems

GPU 기반 스트림 처리 시스템에서의 에너지와 지연시간 공동 인지 관리 기법

목차

Chapter 1. Introduction 1

Chapter 2. Background and Motivation 5
2.1 GPU energy efficiency 5
2.2 Pareto Optimality 6
2.3 Motivation 7

Chapter 3. Design and Implementation 11
3.1 Design Challenges 11
3.2 Overview of Equilibria 12
3.3 Pareto Optima Technique for GPU DVFS 15
3.4 Two-tier Threshold Training Management 22

Chapter 4. Evaluation 25
4.1 Evaluation Setup 25
4.2 Overall Performance Analysis 27
4.3 Performance under various constraints 30
4.4 Comparison with other Pareto Optima Algorithm 31
4.5 Training GPU Power Consumption 33

Chapter 5. Related Work 35
Chapter 6. Conclusion and Future Work 38

Reference 39

more