Moving Towards Autonomous HPC Facilities .
James M. Brandt · 2022
Currently: HPC facility operation and rule-based orchestration is based on human experience, inference, and reaction times.Goal (short term): AI provides data-driven Decision Support for humans to support better resource utilization and higher resource availability (equivalent to level 0 in auto (blind spot warning))Goal (long term): AI used to Orchestrate and Optimize resources and operations subject to physical constraints such as power and priority with humans providing support in terms of repair and iteration on next generation capabilities (equivalent to level 5 in auto (fully autonomous)) SAND2022-12094CThis paper describes objective technical results and analysis.Any subjective views or opinions that might be expressed in the paper do not necessarily represent the views of the U.S.