학술논문

Nanopackaging Chiplets for AI and HPC Applications [feature-3358709] [feature-3358709]
Document Type
Periodical
Source
IEEE Nanotechnology Magazine IEEE Nanotechnology Mag. Nanotechnology Magazine, IEEE. 18(2):21-29 Apr, 2024
Subject
Components, Circuits, Devices and Systems
Computing and Processing
High performance computing
Integrated circuit interconnections
Nanomaterials
Market research
Mechanical factors
Manufacturing
Complexity theory
Nanopackaging
Multichip modules
Packaging
Artificial intelligence
Advanced packaging
chiplets
heterogeneous integration
nanopackaging
Language
ISSN
1932-4510
1942-7808
Abstract
The evolving demands of artificial intelligence (AI) and High-Performance Computing (HPC) applications necessitate dense connectivity between integrated circuit (IC) chips to facilitate high-speed computations. The adoption of chiplet integration through advanced packaging is considered a crucial technology for empowering AI applications which rely on large memory capacities integrated with the compute chip through dense and short interconnects enabling high-bandwidth at low power. These highly integrated systems demand a diverse range of materials to support electrical, mechanical, thermal, and chemical properties. Moreover, these materials must align with packaging processes to ensure compatibility with cost-effective manufacturing solutions. The intricate interplay among various engineering domains adds complexity to the selection of materials, their processability, and compatibility. This article explores the future landscape concerning the material requirements for advanced packaging, examining past and present developments in nanomaterials and nanopackaging techniques that drive this evolving trend.