XBTorch: A unified framework for modeling and co-design of crossbar-based deep learning accelerators
Open Access
- 19 October 2026
- journal article
- Published by AIP Publishing in APL Machine Learning
- Vol. 4 (4)
Abstract
Emerging memory technologies have gained significant attention as a promising pathway to overcome the limitations of conventional computing architectures in deep learning applications. By enabling computation directly within memory, these technologies—built on nanoscale devices with tunable and nonvolatile conductance—offer the potential to drastically reduce energy consumption and latency compared to traditional von Neumann systems. This paper introduces XBTorch (short for CrossBarTorch), a novel simulation framework that integrates seamlessly with PyTorch and provides specialized tools for accurately and efficiently modeling crossbar-based systems based on emerging memory technologies. Through detailed comparisons and case studies involving hardware-aware training and inference, we demonstrate how XBTorch offers a unified interface for key research areas such as device-level modeling, cross-layer co-design, and inference-time fault tolerance. While exemplar studies utilize ferroelectric field-effect transistor models, the framework remains technology-agnostic—supporting other emerging memories, such as resistive RAM, and enabling user-defined custom device models. The code is publicly available at https://github.com/ADAM-Lab-GW/xbtorch.Keywords
Funding Information
- DEVCOM Army Research Laboratory (W911NF-25-2-0073)
- DEVCOM Army Research Laboratory (ECNS21932N)
This publication has 25 references indexed in Scilit:
- A full-stack memristor-based computation-in-memory system with software-hardware co-developmentNature Communications, 2025
- Layer ensemble averaging for fault tolerance in memristive neural networksNature Communications, 2025
- Ferroelectric FDSOI FET modeling for memory and logic applicationsSolid-State Electronics, 2022
- Data-driven RRAM device models using Kriging interpolationScientific Reports, 2022
- Breaking the von Neumann bottleneck: architecture-level processing-in-memory technologyScience China Information Sciences, 2021
- The Future of Memristors: Materials Engineering and Neural NetworksAdvanced Functional Materials, 2020
- Committee machines—a universal method to deal with non-idealities in memristor-based neural networksNature Communications, 2020
- Power-efficient combinatorial optimization using intrinsic noise in memristor Hopfield neural networksNature Electronics, 2020
- Computational phase-change memory: beyond von Neumann computingJournal of Physics D: Applied Physics, 2019
- Neuromorphic computing using non-volatile memoryAdvances in Physics: X, 2016