Impact of Chiplets, Heterogeneous Integration and Modularity on AI and HPC systems

This panel explores the transformative impact of chiplets and heterogeneous integration on AI and HPC, addressing the escalating demands for memory and computing power. Chiplets, discrete components with specialized functions, are reshaping system architectures by enabling large-scale integration and modular designs. This approach optimizes performance, energy efficiency, and scalability in AI applications, from neural networks to data-intensive HPC workloads. Discussions will delve into the benefits of modular architecture, such as enhanced flexibility, easier upgrades, and accelerated innovation cycles. Key topics include design strategies for integrating diverse chiplet functionalities, interoperability standards, and ecosystem development. Attendees will gain insights into how these advancements are redefining the landscape of AI and HPC, fostering collaborative efforts to meet evolving computational challenges effectively. 

Moderator:

John Shalf, Berkeley Laboratory

Panelists:

  • Ramin Farjadrad (Ceo) - Eliyan
  • William Chen (Senior Technical Advisor) - Ase Group
  • Prith Banerjee (Chief Technology Officer) - Ansys, Inc.
  • Dave Ditzel (Founder And CTO) - Esperanto Technologies
  • Francesc Guim - Openchip & Software Technologies