Chiplet interoperability and breakthroughs in AI
By Francisco Soca, Arm and Jonathon Evans, Nvidia
Chiplets offer greater system composability and performance scaling- making them attractive for AI and accelerated computing. To be adopted at scale- chiplets need to be interoperable. Interoperability is achieved through the standardization of chiplet interfaces and protocols.
AMBA CHI C2C is an extension to the widely adopted on-chip CHI protocol for connecting chip(let)-to-chip(let). This presentation shows how AMBA CHI C2C can be used to coherently and securely connect Arm-based systems-on-chip to accelerators over C2C to enable breakthroughs in AI and accelerated computing.Arm and NVIDIA have been collaborating to lay the foundations of this extension. As part of this presentation- NVIDIA will discuss the value of a native AMBA CHI interface as a chiplet interface for accelerator attach.
Related Chiplet
- Bridglets
- Automotive AI Accelerator
- Direct Chiplet Interface
- HBM3e Advanced-packaging chiplet for all workloads
- UCIe AP based 8-bit 170-Gsps Chiplet Transceiver
Related Videos
- Revolutionizing System Design: Impact of Chiplets and Heterogeneous Integration on AI Server
- Impact of Chiplets, Heterogeneous Integration and Modularity on AI and HPC systems
- Paving the Road Ahead: RISC-V and Chiplet Technologies in Modern Automotive and Data Center Architectures
- Balaji Baktha, Paving the Road Ahead RISC V and Chiplet Technologies in Modern Automotive & More
Latest Videos
- Mick Posner talks about the importance of UCIe
- Optimizing Chiplet architectures of RISC-V clusters, GPU and DNN using System-Level IP
- Speeding Up Die-To-Die Interconnectivity
- GENIOEVO™ – Architectural Exploration and Connectivity Management in Advanced Packaging
- Enabling a true open ecosystem for 3D IC design