nanoHEENS: Biomimetic Near-Memory-Computing 16-Core SIMD Processor Node for Evolutive Spiking Neural Networks
Journal
2026 IEEE International Symposium on Circuits and Systems (ISCAS)
Date Issued
2026-05-24
Author(s)
Larre-Alos, Arnau
Vallejo-Mancero, Bernardo
Fernandez, Daniel
Moreno, J. Manuel
Cosp-Vilella, Jordi
Madrenas, Jordi
Type
text::conference output::conference paper not in proceedings
Abstract
The design of a digital spiking neural processor (NP) ASIC for biomimetic evolutive applications is reported. The architecture is briefly described, highlighting its fundamental characteristics, namely: Near-Memory Computing (NMC), efficient Single-Instruction Multiple-Data (SIMD) computation, multi-model support, 16-bit precision, axonal delay support and on-the-fly reconfiguration. A sub-mm2, 16-core NP testchip has been designed in 28 nm CMOS and sent to manufacture. A post-implementation simulation, including a controller and two NPs, is shown to demonstrate the system's correct operation and capabilities