arXiv 2602.08081
Investigating Energy Bounds of Analog Compute-in-Memory with Local Normalization
By Brian Rojkov, Shubham Ranjan, et al.
Published 2026-02-08
Mindmap
Browse the paper's core ideas, clusters, and relationships in a structured outline.
Modern edge AI workloads demand maximum energy efficiency, motivating the pursuit of analog Compute-in-Memory (CIM) architectures. Simultaneously, the popularity of Large-Language-Models (LLMs) drives the adoption of low-bit floating-point formats which prioritize dynamic range. However, the conventional direct-accumulation CIM accommodates floating-points by normalizing them to a shared widened fixed-point scale. C…