Designing quantum chips involves integrating well-established microwave engineering methods with the stringent requirements of ultra-low-temperature quantum physics. This intersection of classical engineering and quantum science makes accurate modeling particularly challenging. To address this, ARTEMIS, developed through the U.S. Department of Energy’s Exascale Computing Project, offers a powerful platform for simulating the intricate electromagnetic interactions inside these devices.

Figure 1. Large-Scale GPU Simulation Reveals Fine Structure of Quantum Chip.

While many quantum chip simulations do not demand extreme computational resources, capturing the fine structural details of this highly complex device required nearly the full capability of Perlmutter. Researchers employed almost all 7,168 NVIDIA GPUs for 24 hours to model a multilayer chip just 10 millimeters wide and 0.3 millimeters thick, featuring etched components as small as one micron. Figure 1 shows Large-Scale GPU Simulation Reveals Fine Structure of Quantum Chip.

Computer-Generated Microchip Etchings

“I’m not aware of anyone who has carried out full physical modeling of microelectronic circuits at the complete system scale of Perlmutter. We were using nearly 7,000 GPUs,” said Nonaka. “The chip was discretized into 11 billion grid cells, and we executed more than a million time steps in just seven hours. That performance enabled us to test three different circuit configurations within a single day. Achieving this turnaround would not have been possible without access to the full system.”

This extraordinary resolution sets the simulation apart. Whereas many approaches simplify chips as “black boxes” due to computational limitations, leveraging Perlmutter’s massively parallel GPU architecture allowed Yao and Nonaka to model the device at a true physical level and reveal its internal operating mechanisms.

“We perform full-wave, physics-based simulations,” Yao explained. “That means we account for the exact materials used on the chip, its layout, the configuration of metal wiring—whether niobium or other metals—the construction of resonators, and their size and shape. Every physical detail is incorporated into the model.”

Beyond its microscopic structural precision, the simulation also replicated laboratory conditions, modeling how qubits interact with one another and with other components of the quantum circuit. Integrating detailed physical chip modeling with real-time simulation capability is what makes this work distinctive, Yao explained. “That combination is essential because we solve partial differential equations—specifically Maxwell’s equations—in the time domain, which allows us to capture nonlinear behavior. Together, these elements give us a truly unique capability.”

NERSC has supported numerous quantum information science efforts through the Quantum Information Science @ Perlmutter program, which allocates Director’s Discretionary Reserve hours on Perlmutter to promising quantum research. Even so, staff noted that undertaking a simulation of this scale was an especially exciting challenge.

“This project stands out as one of the most ambitious quantum initiatives ever run on Perlmutter,” said Katie Klymko, a quantum computing engineer at NERSC. “By leveraging ARTEMIS alongside NERSC’s computational resources, we were able to model quantum hardware features spanning more than four orders of magnitude in scale.”

Source: SciTECHDaily

Cite this article:

Priyadharshini S (2026), Quantum Microchip Simulated in Exceptional Detail Using 7,000 GPUs, AnaTechMaz, pp.447