Beyond Planar Scaling:Fundamental Investigations into 3D Reconfigurable Architectures, Self-Healing Dynamics, and a New Paradigm for Semiconductor Scalin (AFOSR/IOS (SOARD) International Science) T392

Acronym

P_3D-SEMICON

Consortium Coordinator

Carranza De La Cruz, Cesar Alberto

Start Date

2025

End Date

2028

Status

https://purl.org/pe-repo/concytec/estadoProyecto#concluido

Tipo de proyecto

https://purl.org/pe-repo/ocde/tipoProyecto#investigacionBasica

Description

Inspired by previous work on 2D Adaptive Wiring Machines (AWM), which implemented a form of self-healing by continuously forming alternate connection paths, this research extends these ideas into three dimensions. In prior AFOSR-sponsored efforts, we demonstrated that establishing alternative connections allowed continuous self-healing by monitoring switch health and marking defective nodes. Additionally, our experiments revealed a phenomenon we refer to as solitonic behavior—wherein a cellular system dynamically reconfigures its pathways to bypass obstacles, similar to a Terminator reconstituting its connectivity. This proposal seeks to harness and expand these insights within a 3D Adaptive Wiring Panel (AWP) framework. Each cell in the system contains a programmable conductance element—initially configured to short its six neighbors (N-E-W-S-U-D) when in the “1” state, with subsequent versions enabling more flexible interconnections—and integrated programmable heaters to address thermal challenges like dark silicon. The goal is to develop a foundational understanding of 3D scaling, reconfigurability, and self-healing dynamics, potentially extending the semiconductor roadmap and inspiring a revised “3D Moore’s Law.” We aim to establish a 3D scaling framework by extending classical lambda scaling and reformulating Rent’s Rule in a volumetric context, incorporating defect probabilities and connectivity losses at extreme scales. By transitioning from 2D to 3D, we explore the potential for a new scaling paradigm that enables dynamic reconfiguration and alternate path formation for self-healing. Building on prior 2D AWM research, we will investigate how solitonic behavior allows the system to bypass defects and maintain continuous operation. Additionally, we will explore self-assembly, reconfigurability, and thermal management through cellular automata simulations, integrating programmable heaters, and emphasizing software-defined hardware for continuous self-healing and environmental resilience. These concepts will be validated through computational simulations and experimental prototyping, using Monte Carlo methods and 3D-printed macroscale models to demonstrate programmable interconnects, defect tolerance, and solitonic reconfiguration. This research advances a new scaling paradigm by extending self-healing and reconfigurability from 2D to 3D, potentially leading to a revised “3D Moore’s Law.” By integrating continuous self-healing mechanisms such as alternate path formation and solitonic behavior, the system enhances defect tolerance and long-term reliability. It also bridges chiplet strategies with volumetric integration, enabling modular, reconfigurable architectures aligned with industry trends. Furthermore, innovative thermal management via programmable heaters ensures the viability of densely packed 3D systems. Finally, by emphasizing reconfigurability as a core adaptation strategy, this approach addresses fabrication imperfections and operational defects, enhancing overall system resilience.

Keywords

Arquitecturas 3D reconfigurables

;

Semiconductores

;

Auto reparación

;

Escalamiento tecnológico

Área de conocimiento

Engineering and technology

Campo OCDE

https://purl.org/pe-repo/ocde/ford#2.05.01