Nature electronics article on using 3D technologies for scaling neuromorphic systems
In a recent nature electronics article, Melika Payvand from the Insitute of Neuroninformatics (INI) and Elisa Vianello of CEA-LETI write about "Scaling neuromorphic systems with 3D technologies".
In-memory computing has emerged as the solution to the Von Neumann bottleneck in AI accelerators. But large-scale systems need to go beyond a single in-memory computing core to multi-core systems, which re-introduces data movement problems, as data must be moved between multiple cores. The two researchers discuss how various available 3D technologies can be used for scaling in-memory computing, specifically for spike-based neuromorphic hardware.
Link to the article: external page Scaling neuromorphic systems with 3D technologies
Congratulations!