BBCube 3D - Heterogeneous 3D Integration for TB/s Bandwidth in AI and HPC Applications
Open Compute Project via YouTube
Free courses from frontend to fullstack and AI
Google AI Professional Certificate - Learn AI Skills That Get You Hired
Overview
Google, IBM & Meta Certificates — All 10,000+ Courses at 40% Off
One annual plan covers every course and certificate on Coursera. 40% off for a limited time.
Get Full Access
Learn about BBCube 3D technology in this 10-minute conference talk from Hitachi's Norio Chujo at the Open Compute Project. Explore a groundbreaking heterogeneous 3D integration approach that combines bumpless Wafer-on-Wafer and Chip-on-Wafer technologies to achieve unprecedented TB/s bandwidth for AI and HPC applications. Discover how via-last TSVs and ultra-thinning die techniques solve critical cooling and power delivery challenges, enabling the stacking of 47W xPU processors on laminated DRAMs. Understand the implementation of low capacitance TSV technology that delivers 4x higher bandwidth while consuming only one-fifth of the bit access energy compared to HBM2E. Follow the comprehensive presentation covering introduction, background, stacking process, special features, TSV specifications, XPU integration, temperature management, access energy optimization, and audience Q&A.
Syllabus
Introduction
Background
Stacking Process
Spal Features
TSV
XPU
Temperature
Access Energy
Questions
Taught by
Open Compute Project