Impact of precision on neuromorphic algorithm design and costs

Investigate how computational precision requirements affect neuromorphic algorithm design and resource costs by characterizing the implications of precision on neuromorphic computational graphs and performance, including trade-offs when leveraging analog or non-conventional components.

Background

The paper explicitly sets aside precision considerations, noting that precision in digital systems expands computational graphs in ways that are typically manageable. In neuromorphic settings—especially with potential analog or non-standard devices—precision requirements may materially affect algorithm viability and resource consumption.

Understanding precision impacts is therefore essential for evaluating neuromorphic suitability and for guiding hardware–algorithm co-design, but remains unresolved according to the authors.