A 22nm 25.08TOPS/W Multi-Task Transformer Accelerator with Mixed Precision Structured Sparsity and Two-Stage Task-Adaptive Power Management
作者:Zichen Fan, Qirui Zhang, Pierre Abillama, Sara Shoori, Jungho Lee, Chien-Wei Tseng, Wei Meng, Hun-Seok Kim, David Blaauw, Dennis Sylvester · 年份:2025 · DOI:10.23919/vlsitechnologyandcir65189.2025.11075113 · 被引用次数:4 · 研究领域:Advanced DC-DC Converters、Analog and Mixed-Signal Circuit Design、Wireless Power Transfer Systems
This paper presents the first silicon implementation of a transformer accelerator that executes multiple tasks simultaneously, improving per-task efficiency. It applies structured sparsity, multi-precision computation, and a two stage DVFS mechanism yielding highest reported efficiency per task.