Author | zer0
Edit |Desert Shadow
On February 21, NVIDIA announced that the Australian Supercomputing Research Center will use the NVIDIA Cuda Quantum platform, accelerated by the NVIDIA Grace Hopper superchip, at its National Supercomputing and Quantum Computing Innovation Center to run state-of-the-art quantum computing simulations.
Australia's national science agency, the Commonwealth Scientific and Industrial Research Organisation (CSIRO), estimates that quantum computing could bring $2.5 billion worth of opportunities to the Australian market each year, with the potential to create 10,000 new jobs by 2040. To achieve this, quantum computing needs to be introduced to other scientific fields, such as astronomy, life sciences, medicine, finance, etc.
"Whether it's developing algorithms, designing devices, or inventing powerful methods for error correction, calibration, and control, high-performance simulations are critical for researchers to address the grand challenges of quantum computing," said Tim Costa, director of high-performance computing and quantum computing at NVIDIA. Together with the NVIDIA Grace Hopper superchip, Cuda Quantum is helping innovators like the Pawsey Center for Supercomputing to achieve these important breakthroughs and accelerate the adoption of quantum-integrated supercomputing. ”
Supercomputers are the foundation of quantum R&D. The PAWSEY Supercomputing Research Center is committed to providing the NVIDIA Grace Hopper platform to the Australian quantum community and its international partners, and is deploying eight NVIDIA Grace Hopper superchips based on the NVIDIA MGX modular architecture to power the NVIDIA CUDA quantum platform.
Grace Hopper combines ARM-based NVIDIA Grace CPUs with Hopper-based GPUs into a single package and uses NVIDIA NVLlink-C2C technology to interconnect without the need for traditional CPU-GPU PCIe connections, run high-fidelity, scalable quantum simulations on accelerators, and work seamlessly with future quantum hardware infrastructures.
Compared to the latest PCIe technology, the bandwidth between the GPU and the CPU is increased by 7 times. Up to 10x faster application performance for running terabytes of data enables quantum-classical researchers to solve the world's most complex problems.
NVIDIA Cuda Quantum is an open-source hybrid quantum computing platform with rich simulation tools and the ability to program hybrid CPU, GPU, and QPU systems, with the capabilities of hybrid quantum-high-performance computing (HPC) systems. Researchers at the center will use CUDA QUANTUM and NVIDIA Cuquantum, a suite of software development tools optimized to accelerate quantum computing workflows.
The PAWSEY Supercomputing Research Center will deploy the system to run quantum workloads directly on traditional high-performance computing systems, take full advantage of the processing power of these systems, and develop hybrid algorithms that intelligently divide computing into classical and quantum cores to improve computing efficiency through quantum devices. The center will study key areas such as quantum machine learning, chemical simulation, radio astronomy image processing, financial analysis, bioinformatics, and specialized quantum simulators, starting with various quantum variational algorithms.
An example of the use of AI to accelerate quantum computing is that the University of Toronto, St. Jude Children's Research Hospital, NVIDIA have collaborated to develop a new generative pre-trained Transformer (GPT)-based method to compute the ground-state energy of a molecule of interest. Materials science and environmental challenges have opened the door to novel generative quantum algorithms (GQAs).
Mark Stickells, Executive Director of the PAWSEY Supercomputing Research Centre, said: "The research and testbed facilities at the PAWSEY Supercomputing Research Centre are advancing scientific exploration in Australia and around the world. NVIDIA's Cuda Quantum platform will enable our scientists to drive breakthrough innovation in quantum computing research. ”
In the future, useful quantum computing will be valuable in a hybrid form, tightly integrated with supercomputers as accelerators, and co-programmed.