Agentic AI Explores Heterogeneous Chip and Interconnect Architectures (Argonne, UIUC)

Researchers from Argonne National Laboratory and the University of Illinois Urbana-Champaign have developed an agentic artificial intelligence methodology to explore heterogeneous computing architectures. The framework is designed to optimise complex design spaces spanning GPUs, alternative accelerators, and advanced memory interfaces across scaled-up and scaled-out high-performance systems. By automating architectural exploration, the project seeks to accelerate development cycles for next-generation AI and supercomputing platforms.