Researchers at the National Energy Research Scientific Computing Center (NERSC) have successfully simulated a quantum chip in unprecedented detail using the Perlmutter supercomputer. This futuristic simulation operated on the incredible processing power of more than 7,000 NVIDIA GPUs. It massively accelerated our overall understanding of quantum hardware design.
Running a single simulation for the entire 24-hour simulation time on Perlmutter, which is powered by 7,168 NVIDIA GPUs. It was important for the team to carefully document the intricate architecture and operation of a stacked, multi-layered quantum chip. This tiny chip – only 10 millimeters square and 0.3 millimeters thick – features etchings as fine as one micron. For this fine-scale analysis, researchers discretized the chip into 11 billion grid cells. This strategy allowed them to rapidly and accurately prototype thousands of circuit topologies and configurations.
This collaborative endeavor pulled in a variety of groups throughout the larger Berkeley community, including AMCR, QSA, AQT, and NERSC. The invaluable support we received from NERSC was foundational to our success as a fledgling project. Their powerful computational resources and truly staff expertise really made all the difference.
In a reflection on the importance of the simulation, Bert de Jong noted that the simulator is significant because
“This unprecedented simulation, made possible by a broad partnership among scientists and engineers, is a critical step forward to accelerate the design and development of quantum hardware.”
This new simulation let the researchers test three circuit configurations in one day on Perlmutter. During the seven-hour simulation, Perlmutter calculated over a million time steps. This ability to execute long-form simulations at such an accelerated pace made full use of the supercomputer’s capacities.
The researchers combined Maxwell’s equation with the use of partial differential equations in the time domain. This method provided them the means to include nonlinear movement in their models. Yao emphasized the importance of these methods:
“This effort stands out as one of the most ambitious quantum projects on Perlmutter to date, using ARTEMIS and NERSC’s computing capabilities to capture quantum hardware detail over more than four orders of magnitude.”
Yao will be presenting the results of this simulation at the forthcoming International Conference for High Performance Computing, Networking, Storage and Analysis (SC25). We’re in for some thrilling data! Alexandrov’s work highlights an early success from NERSC’s Quantum Information Science @ Perlmutter program. This critical program has been essential in moving the ball forward on a number of different initiatives in quantum information science.
“The combination is instrumental, because we use the partial differential equation, Maxwell’s equation, and we do it in the time domain so we can incorporate nonlinear behavior. All this adds up to give us one-of-a-kind capability.”
Nonaka further explained the level of detail achieved:
“We discretized the chip into 11 billion grid cells. We were able to run over a million time steps in seven hours, which allowed us to evaluate three circuit configurations within a single day on Perlmutter. These simulations would not have been possible in this time frame without the full system.”
Looking ahead, Yao expressed aspirations for future simulations:
“We’d like to do a more quantitative simulation so that we can do a post-process and quantify the spectral behavior of the system.” He also indicated an interest in benchmarking frequency-domain results against other simulations.
The findings from this simulation will be presented by Yao at the upcoming International Conference for High Performance Computing, Networking, Storage, and Analysis (SC25). The work represents another significant contribution from NERSC’s Quantum Information Science @ Perlmutter program, which has been instrumental in supporting various quantum information science initiatives.