The quantum version of CUDA NVIDIA super core star is the god assist of Australian quantum com

Mondo International Updated on 2024-02-23

Author: Mao Shuo.

"NVIDIA wants to support the practical application of quantum computers in the future," Huang said. ”

As a result, the NVIDIA Cuda Quantum platform is being put into use at Australia's National Innovation Centre for Supercomputing and Quantum Computing, further driving quantum computing breakthroughs.

The starting point of all this can be traced back to the GTC Spring Conference in 2023. At that time, NVIDIA announced that it had co-developed the world's first GPU-accelerated quantum computing system, the NVIDIA DGX Quantum, with Quantum Machines.

As a pioneer of a technological revolution, NVIDIA DGX Quantum is not only the world's first GPU-accelerated quantum computing system, but also combines the computing power of NVIDIA Grace Hopper superchips with the flexibility of Cuda Quantum's open-source programming model, and OPX, the world's leading quantum control platform built by Quantum Machines.

Perhaps, when the gears of fate begin to slowly turn, this latest achievement in accelerated computing at this stage is heralding the dawn of a new era in the field of quantum computing.

NVIDIA Cuda Quantum makes the complexity of quantum programming uncountable

In today's fast-moving technology landscape, the launch of NVIDIA Cuda Quantum marks a major breakthrough in the convergence of quantum computing and classical computing. This innovative platform not only expands the programming model of hybrid quantum-classical systems, but also opens up new avenues for quantum computing research and applications through its native support for GPU hybrid computing.

Specifically, NVIDIA Cuda Quantum uses an advanced kernel programming model that supports C++ and Python languages, enabling researchers and developers to write with unprecedented flexibility and efficiency**.

By analogy, NVIDIA Cuda Quantum is powerful when it comes to dealing with massively parallel processing and data-intensive quantum applications. Its GPU pre- and post-processing capabilities, coupled with support for classical optimization algorithms, act like an efficient accelerator that significantly increases the speed and efficiency of computing.

The system-level compiler toolchain introduced by NVIDIA Cuda Quantum brings more empirical capabilities. Its NVQ++ compiler has the "superpower" of decomposing and compiling, which can reduce the complexity of quantum programming to nothing.

With this "superpower", NVIDIA Cuda Quantum is able to create multi-level intermediate representations (MLIR) and quantum intermediate representations (QIR) for quantum cores, which is like building a multi-dimensional world, allowing ** to not only walk freely in different environments, but also to bring more possibilities for future optimization. In fact, this process not only reduces the complexity of programming, but also broadens the space for portability and optimization, making the journey of quantum programming more full of surprises.

Preliminary NVQ++ benchmark results show a staggering 287x end-to-end performance improvement for variational quantum bender-solver (VQE) tasks using NVIDIA CUDA Quantum compared to traditional Pythonic implementations, especially when processing 20-qubit systems, which increases significantly as the system scales.

In addition, Cuda Quantum provides a standard library that covers the basic primitives of quantum algorithms, making it easier for developers to implement complex quantum algorithms.

The platform's interoperability is also part of its robust capabilities, as it not only interacts with partner quantum processing units (QPUs), but also supports QPU emulated through the CuQuantum GPU platform, as well as working with QPU builders to process many different types of qubits. This means that researchers can easily switch between different QPUs, whether analog or physical, as simple as changing the compiler flag.

In the author's opinion, the emergence of NVIDIA Cuda Quantum not only improves the implementation efficiency of quantum algorithms, but also provides a strong impetus for the research and practical application of quantum computing through its high flexibility and scalability. NVIDIA Cuda Quantum's GPU hybrid computing support and system-level compiler toolchain set a new standard for the development of quantum-classical fusion systems, heralding the widespread application of quantum computing technology in more fields in the future, thereby accelerating the transition to quantum advantage and quantum utility.

In fact, researchers in Australia's national supercomputing and quantum computing are using NVIDIA CUDA Quantum, an open-source hybrid quantum computing platform with powerful simulation tools and the ability to program hybrid CPU, GPU and QPU systems, as well as the NVIDIA CUDA QUANTUM software development toolkit, which includes libraries and tools optimized to accelerate quantum computing workflows.

As Mark Stickells, Executive Director of the Pawsey Supercomputing Research Centre, explains: "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. ”

Set up GPUs and CPUs with 7 times the "high speed" and 10 times the performance to build a "big core dirt" for quantum computing

The NVIDIA CUDA Quantum platform at the Pawsey Supercomputing Research Centre in Australia will be accelerated by NVIDIA Grace Hopper superchips at its National Center for Supercomputing and Quantum Computing Innovation, aiming to further push the center to breakthroughs in quantum computing.

Whether it's Oxford Quantum Circuits in Reading, UK, using NVIDIA Grace Hopper in a hybrid QPU GPU system programmed by CUDA Quantum. Or Chicago's Qbraid uses NVIDIA Grace Hopper to build quantum cloud services, or Amsterdam's Fermioniq uses NVIDIA Grace Hopper to develop tensor network algorithms.

Why are superchips frequently "loved" by innovative research in the field of quantum computing?

In fact, the NVIDIA Grace Hopper superchip architecture combines the power of the NVIDIA Hopper GPU with the flexibility and versatility of the NVIDIA Grace CPU. Cleverly "placed" in this one-of-a-kind superchip, the traditional CPU-GPU PCIe connection is forged and the advanced NVIDIA NVLink chip-2-chip (C2C) high-speed channel and NVIDIA NVLink Switch System enable a fast flow of data and memory, seamlessly connecting the worlds of high bandwidth and memory.

NVIDIA Grace Hopper provides an ultra-wide "highway" between GPUs and CPUs compared to the latest PCIe technology, delivering up to 7x more bandwidth and up to 10x higher performance for applications running terabytes of data, enabling researchers in quantum computing to solve the world's most complex problems like never before.

The high-performance superchip is also capable of high-fidelity, scalable quantum simulations on accelerators, and is capable of docking with quantum hardware infrastructure.

Whether developing algorithms, designing devices, or inventing robust methods for error correction, calibration, and control, high-performance simulations are essential for researchers to address the grand challenges in quantum computing. Together with the NVIDIA Grace Hopper superchip, CUDA Quantum is enabling these important breakthroughs at innovative organizations like the Pawsey Center for Supercomputing Research, accelerating the adoption of quantum-integrated supercomputing. Tim Costa, director of high-performance computing and quantum computing at NVIDIA, said.

Currently, the Pawsey Supercomputing Research Center is deploying eight NVIDIA Grace Hopper superchip nodes based on the NVIDIA MGX modular architecture.

With the acceleration of superchips, PAWSEY will be able to run quantum workloads directly on traditional high-performance computing systems, leveraging the processing power of existing systems and improving computational efficiency by developing hybrid algorithms that intelligently divide computing tasks into classical and quantum cores.

Standing at the starting point of the research on "quantum variational algorithms", when the PAWSEY Supercomputing Research Center decided to start deploying it, it heralded the possibility of achieving unprecedented computational efficiency improvements in multiple fields. Quantum machine learning, chemical simulations, radio astronomy image processing, financial analytics, bioinformatics, and research in specialized quantum simulators will all benefit from NVIDIA's powerful computing power. Especially in areas where computationally intensive areas such as chemical simulations and quantum machine learning are extremely computational, this hybrid computing model has the potential to revolutionize the world.

Write at the end

With the acceleration of multi-institution deployment, the overall landscape of quantum computing ecology and development is constantly expanding, and the PAWSEY Supercomputing Research Center is also committed to building bridges for the Australian quantum community and its international partners. And provide it with the NVIDIA Grace Hopper platform.

As the ecological boundaries of quantum computing continue to expand, and today's new generation of large models such as SORA are coming out, their support in quantum computing may mean more efficient data processing and analysis capabilities, especially when dealing with high-complexity tasks. Quantum-accelerated AI algorithms have the potential to significantly improve the speed and efficiency of model training, enabling models to process larger datasets, achieve higher accuracy, and achieve faster inference.

There is reason to believe that quantum computing is poised to become the key to the next generation of AI infrastructure. Although NVIDIA has emphasized that it is not directly involved in quantum computing, its potential contribution to the future of AI infrastructure cannot be ignored, and cross-domain cooperation and technology convergence such as NVIDIA Grace Hopper are expected to accelerate the progress of AI technology and promote the development of global AI applications to a broader and deeper level.

Related Pages