31.5 SoulMate: A 9.8mW Mobile Intelligence System-on-Chip with Mixed-Rank Architecture for On-Device LLM Personalization
Abstract
This work presents SoulMate, a fully on-device mobile intelligence system-on-chip, integrating retrieval-augmented generation (RAG) and fine tuning of a personal LLM. SoulMate is fabricated in 28nm CMOS with a novel mixed-rank token processing and similarity-aware sequence processing architecture. It demonstrates real-time user interaction consumining only 9.8-to-180.5mW power, and state-of-the-art energy efficiency, such as$26.3 \mu \mathrm{J} /$token for inference and$56.8 \mu \mathrm{J} /$token for fine-tuning.
Fields of Study & Subjects
Low-power high-performance VLSI design
Advanced Neural Network Applications
Advanced Memory and Neural Computing
Publication details
- Journal / source
- Digest of technical papers/Digest of technical papers - IEEE International Solid-State Circuits Conference
- Pages
- 540–542
- Publication date
- 15 February 2026
- Language
- EN