AI native Virtual Chiplet Eco-systems: industry best practices to accelerate Chiplet adoption

Matt Holdrege,Research Fellow, Microchip
Marc Meunier, Director HW Ecosystem, Arm
Nandan Nayampally, CCO, Baya Systems
Lichen Weng, SoC Architect, Tenstorrent
Raghu Shankar

Despite the advantages of chiplets, broad adoption addressing a wider range of market opportunities by smaller organizations remains a major challenge.

We need to reduce friction at the front-end of design cycle that includes architecture and business planning thereby increasing design space explorations in multi-vendor environments.

Define AI-native virtual ecosystems built on industry best practices of virtual prototyping that will accelerate chiplet project evaluations.
AI must be applied extensively to increase early-stage explorations while reducing mundane and repetitive tasks. These include semantic and keyword search of specs, rank ordering and scoring top matches, highlighting gaps, code generation, harness to EDA tools, test benches, and benchmarks.

Incorporate best practices including encrypted models, containerized models, fast models, fixed virtual platforms, cloud, local and remote verification IP for interoperability validation, IP libraries, and scenario-based modeling.