Skip to main content
Computing, Environment and Life Sciences

Forward to Exascale: Gauging How Well AI Accelerators Will Work for Science

Argonne is working to understand the effectiveness of new AI accelerator systems for scientific machine learning applications.

Recent advances in hardware have allowed researchers to advance the state of the art in scientific applications, including heterogeneous systems and AI accelerators. For this project, Argonne aims to understand the effectiveness of new AI technologies for scientific applications.

Current hardware and software experiments are aimed at assessing the suitability of AI systems for science and coupling these systems with emerging exascale supercomputers.

The AI testbed center at the Argonne Leadership Computing Facility (ALCF) provides resources for novel AI-hardware systems that will speed the AI components in various applications. For this project, we are developing benchmarks from science applications that help to measure and evaluate performance on diverse AI accelerators. This effort will also help improve the workflow of interconnected experimental and computing facilities.