Watch a 15-minute technical presentation from Arm's Francisco Socal and Nvidia's Jonathon Evans exploring how AMBA CHI C2C protocol enables chiplet interoperability for AI and accelerated computing systems. Learn about extending the established CHI protocol to facilitate chip-to-chip connections, allowing coherent and secure integration between Arm-based systems-on-chip and accelerators. Discover the collaborative efforts between Arm and NVIDIA in developing this extension and understand the benefits of using native AMBA CHI interfaces for accelerator attachment in chiplet-based designs. Gain insights into how standardized chiplet interfaces and protocols are crucial for scaling performance and enabling broader adoption of chiplet technology in AI applications.
Overview
Syllabus
Enabling chiplet interoperability and breakthroughs in AI
Taught by
Open Compute Project
Related Courses
-
Dating and Marrying: An AI Chiplets Perspective
-
Chiplet-based Reconfigurable Open Compute Accelerator Module Architecture and Platform
-
Universal Chiplet Interconnect Express (UCIe): Accelerating the Future of Semiconductor Innovations
-
Advanced Packaging Technologies for Chiplet Adoption and Memory Integration in HPC and AI Applications
-
SoC Construction Using Universal Chiplet Interconnect Express (UCIe) - A Game Changer
-
UCIe - Universal Chiplet Interconnect Express for Die-to-Die Integration
