Reproduce “Monomorphism-based CGRA Mapping via Space and Time Decoupling” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Cristian Tirelli, Rodrigo Otoni, Laura Pozzi
A novel approach to Coarse-Grain Reconfigurable Array (CGRA) mapping decouples the space and time dimensions, using an SMT formulation for temporal scheduling and monomorphism for spatial placement. This method from Università della Svizzera italiana dramatically reduces compilation times by up to 10^5 times for 20x20 CGRAs compared to SAT-MapIt, while retaining equivalent mapping quality.
Key insights
- Explicitly decoupling the complex space and time dimensions of the mapping problem transforms it into two significantly smaller and more manageable search problems.
- Embedding specific capacity and connectivity constraints within the SMT temporal scheduling formulation is sufficient to guarantee the existence of a valid spatial monomorphism.
- Utilizing existing efficient monomorphism extraction algorithms allows for rapid spatial mapping once a valid, spatially-aware temporal schedule is established.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.