Model-Guided Task Mapping on CGRAs
Project Description
Coarse-Grained Reconfigurable Architectures (CGRAs) provide a flexible platform for accelerating computational tasks, but mapping decisions can significantly affect the resulting performance. This project will investigate model-guided approaches for exploring task mapping alternatives on CGRAs. The study will focus on understanding the relationship between computational tasks, CGRA configurations, and mapping choices, and on developing an analytical approach to support the evaluation and selection of task-level mappings.
Supervisor
ZHANG, Wei
Quota
1
Course type
UROP3200
UROP4100
Applicant's Roles
1. Review relevant research on task mapping and performance modeling for CGRAs.
2. Study analytical approaches for evaluating task-level mapping alternatives.
3. Develop and evaluate a prototype framework and summarize the findings in a technical report.
2. Study analytical approaches for evaluating task-level mapping alternatives.
3. Develop and evaluate a prototype framework and summarize the findings in a technical report.
Applicant's Learning Objectives
1. Develop a deeper understanding of task mapping on CGRAs.
2. Gain practical experience in analytical modeling and architecture-aware optimization.
3. Strengthen research, experimental analysis, and technical communication skills.
2. Gain practical experience in analytical modeling and architecture-aware optimization.
3. Strengthen research, experimental analysis, and technical communication skills.
Complexity of the project
Moderate