NVIDIA is replicating its successful experience in AI to quantum computing

This article is reproduced from Lei Feng.com. If you need to reprint, please go to the official website of Lei Feng.com to apply for authorization.

To some people, quantum computing (Quantum computer) may sound like science fiction, a scenario decades away.

In fact, many people around the world have already invested in this cutting-edge computing research. There are more than 2,100 quantum computing research papers published, more than 250 quantum computing start-up companies, and 22 National policies related to quantum computing.

Quantum computing is a new computing model that follows the laws of quantum mechanics to regulate quantum information units for calculation. It is usually compared with classical computing. . From a principle point of view, quantum computing can have calculation speeds faster than classical computing, and this gap may be as high as a trillion times.

Quantum computing is expected to overcome many challenges faced today and promote the development of various tasks from drug research and development to weather forecasting, and can play a huge role in future HPC. Because of this, a large number of companies and researchers are investing resources into studying quantum computing.

Currently, there are many options for physical platforms to achieve quantum computing, such as superconductors, ion traps, neutral atoms, silicon quantum, light quantum, etc. However, they all face different challenges. .

To accelerate the development of quantum computing, Hybrid quantum computing is expected to realize the first practical applications of quantum computing.

The so-called hybrid quantum computing means that quantum computers and classical computers work together to give full play to the advantages of classical computing (such as CPU and GPU) in traditional operations, such as circuit optimization, correction and error correction, and Advantages of system-level quantum processors (QPUs) as new accelerators.

Compared with CPU, GPU is a good choice for hybrid quantum computing, because GPU can shorten the execution time of traditional jobs and greatly reduce the communication delay between classical computers and quantum computers. , which is the main bottleneck facing today’s hybrid quantum operations.

Meanwhile, another big challenge is software tools. As an emerging piece of hardware, quantum processors need to be programmed to realize their value. Researchers can only use quantum equivalent to low-level assembly code. In other words, only quantum computing experts can program quantum accelerators. #This also makes it difficult to promote the rapid development of quantum computing. Therefore,

The field of quantum computing requires a unified programming model and compiler tool chain.

The compiler allows scientists to easily port part of their HPC applications to a simulated QPU and then to a real QPU, efficiently finding ways to accelerate quantum computing work.

With GPU-accelerated simulation tools, programming models, and compiler toolchains all brought together, HPC researchers can begin building the hybrid quantum data center of the future.

Nvidia, which has industry-leading high-performance GPUs and extensive experience in HPC and AI, can help it quickly establish unique products and advantages in the field of quantum computing.

Nvidia has indeed begun to copy its successful experience in the field of AI to the field of quantum computing. Starting from the latest software for developers, lowering the threshold for developers to use it, helping developers in the field of quantum computing solve problems and create value.

Once quantum computing researchers and users choose NVIDIA's tools, they will naturally It can help Nvidia seize the opportunity in the field of quantum computing.

At GTC 2021, NVIDIA announced the launch of cuQuantum SDK, which aims to accelerate quantum circuit simulations running on GPUs. Today, dozens of quantum organizations are already using the cuQuantum software development kit to accelerate their quantum circuit simulations on GPUs.

Recently, AWS provided cuQuantum in the Braket service and demonstrated that

cuQuantum achieved 900 times acceleration on quantum machine learning workloads while reducing costs by 3.5 times.

Another important value of cuQuantum in promoting the development of quantum computing lies in its ability to implement accelerated computing on major quantum software frameworks, including Google's qsim and IBM's Qiskit Aer , Xanadu’s PennyLane and Classiq’s Quantum Algorithm Design platform.

For scientists and developers, users of these frameworks can access GPU acceleration without any further coding. For Nvidia, it will mean its important value in the quantum computing software framework, as well as giving full play to the role of its GPU in hybrid quantum computing.

On July 12, 2022, NVIDIA continued to move forward in the field of quantum computing and released QODA, a unified computing platform.

The goal of Quantum Optimized Device Architecture (QODA) is to make quantum computing more accessible by creating a coherent hybrid quantum classical programming model. QODA also enables experts in the HPC and AI fields to easily port their applications to public clouds, NVIDIA DGX systems, or supercomputing centers equipped with a large number of NVIDIA GPUs.

For quantum organizations already using the cuQuantum software development kit, NVIDIA QODA enables developers to build complete quantum applications that can be simulated on GPU-accelerated supercomputers with NVIDIA cuQuantum .

Like AI and high-performance computing, ecology is the key to success, so software and hardware partners are crucial to NVIDIA's success in the field of quantum computing.

Q2B 22 At the Tokyo Quantum Computing Conference, Nvidia announced partnerships with quantum hardware vendors IQM quantum Computers, Pasqal, Quantum, Quantum Brilliance and Xanadu, software vendors QC Ware and Zapata Computing, and supercomputing centers The German Jurich Research Center, Lawrence Berkeley National Laboratory and Oak Ridge National Laboratory are cooperating on QODA.

NVIDIA CEO Jensen Huang has always emphasized that what NVIDIA needs to do is to create new products and markets, rather than seize existing markets. Quantum computing is such a brand-new market. Nvidia’s choice of technical route and entry point in the field of quantum computing will help it seize the opportunity of quantum computing.

But we must also note that quantum computing still has a long way to go, and it is still difficult to determine who can have quantum hegemony.

The above is the detailed content of NVIDIA is replicating its successful experience in AI to quantum computing. For more information, please follow other related articles on the PHP Chinese website!