CELLA: A 28nm Compute-Memory Co-Optimized Real-Time Digital CIM-based Edge LLM Accelerator with 1.78ms-response in Prefill and 31.32 Token/s in Decoding

Z. Wu, Z. He, B. Yang, Y. Wang, S. Wei, Y. Hu, F. Tu, S. Yin

January 2025

Abstract

Co-Corresponding Author

Type

Conference paper

Publication

Symposium on VLSI Technology and Circuits (VLSI), Kyoto, Japan