Overview
- The team leveraged nearly all 7,168 NVIDIA GPUs on NERSC’s Perlmutter, discretizing a 10 mm by 10 mm chip into 11 billion grid cells.
- More than a million time steps completed in about seven hours enabled evaluation of three circuit configurations within a single day.
- The simulation used physics-grounded, time-domain full-wave modeling that captured materials, wiring, resonator geometry, and nonlinear electromagnetic interactions.
- Researchers from Berkeley Lab and UC Berkeley led the effort with QSA and AQT, modeling a device designed by Irfan Siddiqi’s Quantum Nanoelectronics Laboratory.
- NERSC engineers called it one of Perlmutter’s most ambitious quantum projects, and the team will next conduct quantitative, frequency-domain studies and compare results with a fabricated chip, with highlights slated for SC25.