Reproduce “An Efficient Multicast Addressing Encoding Scheme for Multi-Core Neuromorphic Processors” 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
Zhe Su, Aron Bencsik, Giacomo Indiveri, Davide Bertozzi
The authors introduce HBS, a hierarchical bit-mask encoding for multicast destinations in neuromorphic networks. It selects subgroups at successive tree levels, preserving symbol-based routing-bit scaling while representing more destination combinations. In a synthesized and simulated 16-core network using navigation-task spike traces, HBS reduced area by 9.4% and total modeled energy by 49.1% versus symbol-based encoding; these results are setup-specific.
Key insights
- HBS encoding offers a more flexible multicast addressing capability than symbol-based schemes, allowing for a greater variety of destination core combinations without significant routing bit overhead.
- The hierarchical nature of HBS significantly reduces the occurrence of "illegal events" that need to be filtered at destination cores, leading to energy savings compared to region-based routing.
- Optimal parameter k selection and alignment with hierarchical Network-on-Chip topologies enable a balance between addressing flexibility and efficient communication overhead, improving scalability.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.