Learning-Based Dynamic Pinning of Parallelized Applications in Many-Core Systems

Keyphrases

• Adaptation 100%
• Parallelizing 100%
• Learning-based 100%
• Non-uniform Memory Access 100%
• Many-core Systems 100%
• Access Nodes 66%
• Multi-objective Criteria 66%
• Performance Improvement 33%
• Resource Allocation 33%
• Resource Availability 33%
• Two-level 33%
• Limited Availability 33%
• Adaptive Rate 33%
• Low Switching Frequency 33%
• Rapid Variation 33%
• Noise Measurement 33%
• Parallel Applications 33%
• Learning-based Framework 33%
• Performance Variation 33%
• Scheduler 33%
• Memory Access 33%
• Irregular Memory Access Patterns 33%
• Learning Dynamics 33%
• Linux Kernel 33%
• High-frequency Switching 33%
• Node Level 33%
• Learning Scheme 33%
• Noise Variation 33%
• Dynamic Placement 33%
• Instructions per Second 33%

Computer Science

• Multiobjective 100%
• Level Adaptation 100%
• Many-Core 100%
• Non-Uniform Memory Access 100%
• Switching Frequency 100%
• Performance Improvement 50%
• Parallel Application 50%
• Non-Uniform Memory Access Architecture 50%
• Memory Access 50%
• Learning Scheme 50%
• Memory Access Pattern 50%
• Measurement Noise 50%
• Performance Variation 50%
• Parallel Process 50%
• Resource Allocation 50%