04/28/2025, 05:14 AM UTC
2025年IEEE电子元件与技术会议（ECTC）讨论主题ECTC discussion themes
➀ AMD高级副总裁Samuel Naffziger探讨AI时代对泽塔级高效计算的需求与解决方案，强调从设备到数据中心的系统性创新；
➁ 英特尔提出基于毛细管作用的流体自对准技术，通过液滴自扩展实现3D芯片堆叠，亚微米级对准精度提升混合键合工艺；
➂ IBM展示机器学习模型将布线热阻预测误差降至<15%，相比传统方法（误差300%）大幅提升，可优化先进封装散热可靠性。
➀ AMD Senior VP Samuel Naffziger discusses the demand and solutions for efficient zettascale computing in the AI era, emphasizing holistic innovation across devices to datacenters;
➁ Intel introduces fluidic self-alignment technology using capillary action for precise 3D chip stacking in hybrid bonding, achieving sub-micron alignment accuracy;
➂ IBM demonstrates machine learning models reducing thermal resistance prediction errors to <15%, significantly outperforming traditional methods (300% error) for advanced packaging reliability.
---
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）。
