DS-CIM: Digital Stochastic Computing-In-Memory Featuring Accurate OR-Accumulation via Sample Region Remapping for Edge AI Models
DS-CIM:数字随机计算-内存通过样本区域重映射实现准确的或-累积用于边缘AI模型
机构 * The Hong Kong University of Science and Technology(香港科学与技术大学) ; South China University of Technology(华南理工大学) ; MoE Engineering Research Center of Design and Technology Co-Optimization of IC(MoE设计与技术协同优化集成电路工程研究中心) ; Westlake University(西湖大学)
AI总结 DS-CIM通过样本区域重映射实现高精度和高效率的边缘AI模型计算,适用于INT8 ResNet18和FP8 LLaMA-7B模型。
Comments Accepted by 2026 Design, Automation and Test in Europe Conference (DATE)