05/22/2025, 07:10 AM UTC
利用创新铁电技术开发突破性高效能人工智能Development of Groundbreaking and Energy-Efficient AI with Innovative Ferroelectric Technology
➀ Fraunhofer IPMS联合欧韩合作伙伴主导的ViTFOX项目，利用铁电氧化物开发视觉转换器（ViT）架构，目标实现适用于边缘AI的超过50 TOPS/W能效；
➁ 通过存内计算技术整合数据处理与存储，降低延迟和能耗，采用与硅基器件兼容的铪锆氧化物（HZO）等材料；
➂ 跨国合作推动半导体创新，结合欧洲在铁电隧道结与韩国3D FeRAM的技术优势，促进硬件-软件协同优化平台发展。
➀ The ViTFOX project, led by Fraunhofer IPMS and European-Korean partners, develops a Vision-Transformer (ViT) architecture using ferroelectric oxides to achieve AI energy efficiency exceeding 50 TOPS/W for edge applications;
➁ The project integrates computing-in-memory technology to reduce latency and energy consumption, leveraging ferroelectric materials like HZO compatible with silicon-based devices;
➂ Cross-border collaboration aims to advance semiconductor innovation, combining European expertise in epitaxial ferroelectric junctions with Korean 3D FeRAM development to enhance hardware-software co-optimization.
---
本文由大语言模型（LLM）生成，旨在为读者提供半导体新闻内容的知识扩展（Beta）。
