About this document
Heterogeneous NPUs with PCM AIMC by 20307130063fudan is a document available to read on EtoBox.
The document presents a heterogeneous embedded Neural Processing Unit (NPU) architecture designed for deep learning inference, leveraging Phase-Change Memory (PCM) for Analog In-Memory Computing (AIMC) to meet the energy, area, and cost constraints of edge AI applications. This architecture integrates both digital and analog processing nodes to enhance computational efficiency and flexibility, supporting various neural network models including transformers. The proposed NPU is anticipated to deliver competi
- Author
- 20307130063fudan
- Language
- EN