Specialists from the Berkeley laboratory conducted an exceptional simulation using over 7,000 NVIDIA graphics processors to develop next-generation chip technology. This was reported by UNN with reference to Lawrence Berkeley National Laboratory and Phys.
Details
Specialists from the Lawrence Berkeley National Laboratory (Berkeley Lab) and the University of California, Berkeley, conducted a study simulating next-generation quantum chips. To approach new technology and create better quantum equipment, electromagnetic models developed by laboratory specialists are used. They are being improved by Quantum Systems Accelerator (QSA) researchers Zhi Jackie Yao and Andy Nonaka from the Applied Mathematics and Computational Research (AMCR) Division.
The computational model predicts how design decisions affect the propagation of electromagnetic waves in the chip. .. To ensure proper signal coupling and avoid unwanted crosstalk
Researchers used the exascale simulation tool ARTEMIS to model and optimize a chip developed in collaboration with Irfan Siddiqi's Quantum Nanoelectronics Laboratory at the University of California, Berkeley, and Berkeley Lab's Advanced Quantum Testbed (AQT). This work will be presented in Yao's technical demonstration at the International Conference for High Performance Computing, Networking, Storage, and Analysis (SC25).
Quantum chip design involves traditional microwave engineering in addition to advanced low-temperature physics. This makes a classical electromagnetic modeling tool such as ARTEMIS important for the project's implementation.
Reference
ARTEMIS, which was developed as part of the U.S. Department of Energy's "Exascale Computing Project" initiative.
Thousands of NVIDIA GPUs for Quantizing Billions of Cells
Simulating the fine details of this tiny, extremely complex chip required almost all the power. Researchers used almost all 7,168 NVIDIA GPUs for 24 hours to capture the structure and functions of a multilayer chip measuring only 10 square millimeters and 0.3 millimeters thick, with engravings only one micron wide.
I don't know anyone who has ever done physical modeling of microelectronic circuits at the full scale of the Perlmutter system. We used almost 7,000 GPUs
We discretized the chip into 11 billion grid cells. We were able to perform over a million time steps in seven hours, which allowed us to evaluate three circuit configurations in one day on Perlmutter. These simulations would not have been possible in this time frame without the full system
According to the specialist's colleague, Jackie Yao, the scientists conducted a full-wave physical simulation. "That is, it is important to us what material you use on the chip, its layout, how you connect the metal — niobium or other metal wires — how you build the resonators, what size, what shape, what material you use."
These physical details are important to us, and we include them in our model
In addition to a detailed overview of the chip, the simulation mimicked the experience of laboratory experiments — how qubits interact with each other and with other parts of the quantum circuit.
The combination of these qualities — focusing on the physical design of the chip and the ability to simulate in real time — is part of what made the simulation unique.
This combination is instrumental because we use a partial differential equation, Maxwell's equations, and we do it in the time domain so that we can account for nonlinear behavior. All of this together gives us unique capabilities
According to Katie Klimko, a NERSC quantum computing engineer who worked on the project, the implemented initiative is one of the most ambitious quantum projects to date, using the computational capabilities of ARTEMIS and NERSC to capture the details of quantum hardware over four orders of magnitude.
Recall
Valve co-founder and CEO Gabe Newell is developing a chip that will allow the human brain to interact more closely with a personal computer. It is clarified that this is not yet about creating a full-fledged implant.