Publication
Enabling Computing-in-Memory with Conventional and Emerging Device Technologies
Sutirtha Bhattacharyya; Mohamed Nadeem; Sajjad Parvin; Sallar Ahmadi-Pour; Rolf Drechsler; Chandan Jha
In: 2nd International Conference on Intelligent Computing and Systems at the Edge. International Conference on Intelligent Computing and Systems at the Edge (ICEdge-2026), December 17-19, Bengaluru, India, IEEE, 2026.
Abstract
Compute-in-Memory (CIM) is an emerging paradigm that addresses the limitations of the memory bottleneck in conventional computing architectures by enabling computation within memory structures. Unlike traditional ASIC design flows, CIM requires novel design automation methodologies spanning device, circuit, architecture, and system levels. This work provides an overview of recent advances in CIM, with emphasis on developments in device technologies, circuit design methodologies, synthesis techniques, and formal verification frameworks.
This work describes recent advancements in CIM using three memory technologies: Static Random-Access Memory (SRAM), Dynamic RandomAccess Memory (DRAM), and Resistive Random-Access Memory (RRAM). In this paper, we highlight the similarities in CIM at the higher level of abstraction, especially in synthesis and verification, and call for technologyindependent automated flows. We believe that a technology-independent framework can help the transferability of synthesis and verification techniques across different technologies. This will lead to the identification of common open challenges and will promote standardization and provide a common basis for evaluating and developing methodologies across
different technological platforms.