AI Insight
Researchers have developed CIM-Explorer, a modular software toolkit designed to optimize Binary Neural Networks (BNNs) and Ternary Neural Networks (TNNs) for implementation on Resistive Random Access Memory (RRAM) crossbars used in Computing-in-Memory architectures. The toolkit addresses limitations of existing software by integrating compilation, simulation, and Design Space Exploration capabilities in a single end-to-end framework, specifically tailored for binary and ternary quantization rather than traditional 8-bit quantization. CIM-Explorer enables researchers to estimate accuracy across different crossbar parameters and mappings, facilitating the entire design process from early parameter selection to final chip compilation.
Why it matters
This tool could accelerate the development of energy-efficient AI hardware by overcoming the von Neumann bottleneck, which limits data transfer between memory and processors. By providing comprehensive design exploration capabilities for RRAM-based neural network implementations, CIM-Explorer may enable more practical deployment of low-power AI systems in edge computing and IoT applications.
Understand the Science
arXiv:2505.14303v3 Announce Type: replace-cross
Abstract: Using Resistive Random Access Memory (RRAM) crossbars in Computing-in-Memory (CIM) architectures offers a promising solution to overcome the von Neumann bottleneck. Due to non-idealities like cell variability, RRAM crossbars are often operated in binary mode, utilizing only two states: Low Resistive State (LRS) and High Resistive State (HRS). Binary Neural Networks (BNNs) and Ternary Neural Networks (TNNs) are well-suited for this hardware due to their efficient mapping. Existing software projects for RRAM-based CIM typically focus on only one aspect: compilation, simulation, or Design Space Exploration (DSE). Moreover, they often rely on classical 8 bit quantization. To address these limitations, we introduce CIM-Explorer, a modular toolkit for optimizing BNN and TNN inference on RRAM crossbars. CIM-Explorer includes an end-to-end compiler stack, multiple mapping options, and simulators, enabling a DSE flow for accuracy estimation across different crossbar parameters and mappings. CIM-Explorer can accompany the entire design process, from early accuracy estimation for specific crossbar parameters, to selecting an appropriate mapping, and compiling BNNs and TNNs for a finalized crossbar chip. In DSE case studies, we demonstrate the expected accuracy for various mappings and crossbar parameters. CIM-Explorer can be found on GitHub.
Source: Optimizing Binary and Ternary Neural Network Inference on RRAM Crossbars using CIM-Explorer