Speaker Presentations
Day 1: August 13
- Architectures for the AI Era, from a TPU to the Extreme
Norm Jouppi (Google) - Chakra + ASTRA-sim: An open ecosystem for advancing benchmarking and co-design for diverse AI supercomputers
Tushar Krishna (Georgia Institute of Technology) - ModSim Challenges in Secure and Resilient AI System Design
Pradip Bose (IBM) - Performance Modeling and System Design Insights for AI Foundation Models
Shashank Subramanian (Lawrence Berkeley National Laboratory) - PerfVec: Generalizable Representation Learning for Performance and Energy Modeling of Computer Systems
Lingda Li (Brookhaven National Laboratory) - Hierarchical Modeling and Exploration of Large-Scale Superconducting Supercomputer Architectures
Andreas Gerstlauer (University of Texas at Austin) - Is Neuromorphic Computing Ready for Prime Time?
Brad Aimone (Sandia National Laboratories) - Power-Aware Performance: Modeling for the Physical Limits of Extreme Computing
Marco Scrbak (AMD)
Day 2: August 14
- A National Security AI Platform: driving accelerated science with AI
Timo Bremer (Lawrence Livermore National Laboratory) - Integrating generative AI, multi-scale modeling and simulations with experiments-in-the-loop for biotherapeutic design
Arvind Ramanathan (Argonne National Laboratory)
Day 3: August 15
- Sub-pJ/bit Optical Connectivity for AI Clusters
Keren Bergman (Columbia University) - Post Exascale: Workloads and Emerging Trends
Heidi Poxon (NVIDIA) - Zetta-what and Zetta-how: What comes next after Exascale?
Larry Kaplan (Hewlett Packard Enterprise)