상세 보기
Lupin: Spatial Resource Stealing with Outlier-First Encoding for Mixed-Precision LLM Acceleration
- Kim, Taein;
- Han, Sukhyun;
- Kim, Seongwook;
- Byeon, Gwangeun;
- Lee, Jungmin;
- ... Hong, Seokin
Citations
SCOPUS
0초록
LLM inference often exceeds on-chip memory capacity, causing frequent external memory access. Quantization reduces memory cost but loses accuracy due to outliers. Prior mixed-precision accelerators address this issue with encoding schemes, but often result in accuracy degradation for LLMs and pipeline stalls. We present Lupin, an algorithm-architecture co-design with Outlier-First Encoding, which stores outliers in high precision by reallocating less critical normal values. This preserves maximal outlier representation and enables stall-free execution with low-precision MAC units. Experiments show that Lupin maintains accuracy while achieving a 2.02× speedup.
키워드
Large language models; Mixed-precision; Outlier; Quantization
- 제목
- Lupin: Spatial Resource Stealing with Outlier-First Encoding for Mixed-Precision LLM Acceleration
- 저자
- Kim, Taein; Han, Sukhyun; Kim, Seongwook; Byeon, Gwangeun; Lee, Jungmin; Hong, Seokin
- 발행일
- 2026
- 유형
- Conference Paper
- 저널명
- Proceedings -Design, Automation and Test in Europe, DATE
- 언어
- ENG
- 출판사
- Institute of Electrical and Electronics Engineers Inc.
- ISSN
- P 1530-1591