Should TinyML Processors Be Multi-core - Exploring Edge Computing Architecture

Explore a thought-provoking keynote presentation from tinyML EMEA where Associate Professor Marian Verhelst from KU Leuven examines the evolution and future of tinyML processor architectures. Delve into the critical considerations surrounding multi-core implementations in edge computing, analyzing how real-time deployment of tinyML algorithms impacts smart spaces, augmented humans, and personalized healthcare applications. Learn about the progression from single-core designs to scaled-up datapath arrays and multi-core implementations, while examining the trade-offs between power budgets, efficiency, and throughput. Investigate key questions about homogeneous versus heterogeneous multi-core designs and workload scheduling across edge SoCs, gaining insights into the challenges and opportunities that lie ahead in the field of tiny machine learning processors.