GPU power capping on Perlmutter @ NERSC
Power is a crucial constraint in supercomputing, especially as systems scale to exascale levels. Supercomputers must operate efficiently within strict power limits in order to continue to advance scientific computing. Application-aware power management offers an effective solution to this challenge with a slight performance degradation. In this talk, we will introduce the newly enabled GPU power-capping features on Perlmutter, sharing experimental results from several top applications at NERSC. We encourage users to explore these options on Perlmutter using their own workloads.
Quantum Computing Access @ NERSC (QCAN)
Quantum hardware will likely be an important part of future NERSC systems. In preparation for this, NERSC has recently started offering access to a number of quantum hardware technologies, including neutral atoms and superconducting qubits, through partnerships with QuEra Computing and IBM Quantum. In this talk, we will give an overview of how these technologies work, what running applications look like, and how to gain access to these systems.
Agenda: February 20
| Time | Topic | Presenters |
|---|---|---|
| 11 a.m. | NERSC updates, calls for participation, future training events | |
| GPU power-capping at NERSC | Zhenji Zhao, NERSC | |
| Quantum Computing Access @ NERSC (QCAN) | Katie Klymko and Daan Camps, NERSC |