For decades, scientists and engineers have anticipated that the exponential scaling of conventional von Neumann computers would eventually plateau, struggling to meet the ever-growing computational demands of modern applications. While von Neumann architectures remain the foundation of general-purpose computing, alternative “unconventional” computing paradigms have demonstrated a superior performance in specific targeted domains.
One such paradigm, the Cellular Nonlinear Network (CellNN), was first proposed by Prof. Leon Chua and inspired by the structural and functional principles of the human brain. CellNNs harness the local interactions of large arrays of analog processing units, enabling massively parallel operations with impressive efficiency. At their peak, CellNN-based hardware systems achieved frame processing rates surpassing 1,000 frames per second with excellent energy efficiency, earning them the name “kilo-fps processors”. Despite these achievements, hardware scalability posed a significant challenge since each analog processing unit required a large footprint to ensure the necessary flexibility to execute diverse tasks, thus limiting the overall network size.
Recent developments in memristor devices have reignited this interest in CellNNs, conjuring the image of a reborn phoenix rising from the ashes. This resurgence holds the promise of more compact and efficient analog electronics that may once again position CellNNs at the forefront of ultra-fast image processing and, more broadly, signal processing applications. In this talk, I will explore how these emerging devices breathe new life into CellNN hardware, highlighting improvements in the design flexibility, area efficiency, and computational capability. I will also review the current progress in memristor-based CellNN implementations and share insights into how this technology propels CellNNs toward significant resurgence. Finally, I will discuss promising opportunities on the horizon, illustrating how the rebirth of CellNNs may soon reshape both scientific research and industry innovations alike.
