Computing-in-memory (CIM) is renowned in deep learning due to its high energy efficiency resulting from highly parallel computing with minimal data movement. However, current SRAM-based CIM designs suffer from long latency for loading weight or feature maps from DRAM for large AI models. Moreover, previous SRAM-based CIM architectures lack end-to-end model inference. To address these issues, this paper proposes CIMR-V, an end-to-end CIM accelerator with RISC-V that incorporates CIM layer fusion, convolution/max pooling pipeline, and weight fusion, resulting in an 85.14\% reduction in latency for the keyword spotting model. Furthermore, the proposed CIM-type instructions facilitate end-to-end AI model inference and full stack flow, effectively synergizing the high energy efficiency of CIM and the high programmability of RISC-V. Implemented using TSMC 28nm technology, the proposed design achieves an energy efficiency of 3707.84 TOPS/W and 26.21 TOPS at 50 MHz.
@article{arxiv.2503.22072,
title = {CIMR-V: An End-to-End SRAM-based CIM Accelerator with RISC-V for AI Edge Device},
author = {Yan-Cheng Guo and and Tian-Sheuan Chang and Chih-Sheng Lin and Bo-Cheng Chiou and Chih-Ming Lai and Shyh-Shyuan Sheu and Wei-Chung Lo and Shih-Chieh Chang},
journal= {arXiv preprint arXiv:2503.22072},
year = {2025}
}
Comments
published in IEEE International Symposium on Circuits and Systems (IEEE ISCAS 2024)