The race for quantum computing is on.

Based on quantum theory and following the laws of quantum mechanics, the nascent technology has been touted for its capability to solve deeper, more complex problems than traditional computing – and in a fraction of the time.

But quantum's potential has yet to be fully realized, with many experts saying that its real-world application is still a long way off – perhaps decades.

To help hasten its development and support researchers and developers, tech giant Nvidia announced at its Nvidia GTC event DGX Quantum. Built with Israeli startup Quantum Machines, the system is designed to provide a new architecture for researchers working in high-performance and low-latency quantum classical computing.

“Quantum-accelerated supercomputing has the potential to reshape science and industry with capabilities that can serve humanity in enormous ways,” said Tim Costa, director of high-performance computing (HPC) and quantum at Nvidia.

DGX Quantum, he said, will “enable researchers to push the boundaries of quantum-classical computing.”

Critical Challenges for Quantum to Overcome

Nvidia – along with IBM, Microsoft, Intel, and others – has been making significant investments in quantum, which Costa described in a session at last fall’s GTC event as “a fundamentally new model of computing.”

The company is positioning itself as a leader in a market projected to grow from $712.2 million in 2022, to $4.758 billion by 2029, representing a compound annual growth rate (CAGR) of 31.2%.

But while there are exciting possibilities in the technology, there are unique challenges, as well.

“Quantum computing will require tight integration with classical high performance computing to realize its potential,” Sam Stanwyck, Nvidia quantum computing product lead, said in a statement to SDxCentral.

He pointed to a recent blog from Microsoft that estimated that classical compute at the petascale level is required for fault-tolerant quantum computing. This is because all quantum applications are hybrid, he explained, and many of the subroutines required for quantum computing are compute-intensive classical workloads – such as quantum error correction, calibration, and control. All of these can be dramatically accelerated by GPUs.

“It is also critical that the quantum computer and the classical computer be very tightly coupled so as to not create bottlenecks and to avoid decay of the quantum processor,” Stanwyck said.

Quantum error correction is the best example of this, he said. It is essential to realize a fault-tolerant quantum computer and requires high-performance classical processing that typically must occur within microseconds.

“There is no solution out there today that connects quantum processors with high-performance classical processors with latencies that low,” Stanwyck said, and therefore useful quantum error correction is very challenging.

Large Scale, Accelerated Error Correction

But DGX Quantum “represents a huge leap forward in this regard,” Stanwyck said.

Nvidia calls it the world’s first GPU-accelerated quantum computing system, providing 400-nanosecond latency between the quantum control system and the GPU.

“This will enable large-scale accelerated experiments with error correction for the first time,” Stanwyck said.

DGX Quantum brings together Nvidia’s accelerated computing platform – which is enabled by its Grace Hopper Superchip and CUDA Quantum open-source programming model – with Quantum Machines’ advanced quantum control platform OPX.

This will allow for sub-microsecond latency between GPUs and quantum processing units (QPUs). These, also known as quantum processors, serve as the “brain” of a quantum computer, according to Nvidia. They use the behaviors of particles such as electrons or photons to perform certain kinds of calculations quicker than the processors in today’s computers.

Nvidia co-Founder, President, and CEO Jensen Huang explained in his GTC keynote yesterday that error correction on a large number of qubits is necessary to recover data from quantum noise and decoherence. These are two major obstacles to the large-scale quantum computing implementation.

The quantum control link developed in partnership with Quantum Machines connects Nvidia GPUs to a quantum computer “to do error correction at extremely high speeds,” he said.

The system’s underpinning Grace Hopper is “supercharged” for “giant-scale” AI and HPC applications, delivering up to 10-times higher performance for applications running terabytes of data, according to Nvidia.

And, Quantum Machines’ OPX universal quantum control system brings real-time classical compute engines to the heart of the quantum control stack to maximize performance of any QPU and open new possibilities in quantum algorithms.

Both Grace Hopper and OPX can be scaled to fit the size of the system, from a few qubit QPU to a quantum-accelerated supercomputer.

Integrating Quantum and Classical for a New Generation of Innovators

The new system will allow researchers to build “extraordinarily powerful” applications combining quantum computing with state-of-the-art classical computing – thus enabling calibration, control, quantum error correction and hybrid algorithms, Nvidia says.

“More broadly, Nvidia is building the platform to enable the required integration of quantum and classical systems,” Stanwyck said.

Isaid tamar Sivan, Quantum Machines co-Founder and CEO, commented that: “We are heading toward a new age of quantum computing that is more accessible to more researchers than ever.”

The new collaboration will “enable a new generation of innovators to solve some of the world’s greatest challenges,” he said.

Ultimately, Huang described an overarching partnership between Nvidia and the global quantum computing research community; the company’s quantum platform consists of libraries and systems for researchers to advance quantum models, system architectures and algorithms.

“Although commercial quantum computers are still a decade or two away, we're delighted to support this large and vibrant research community,” Huang said.

CUDA Now Open Source

In addition to DGX Quantum, Nvidia today announced the open-source release of the CUDA Quantum programming model.

Stanwyck explained that CUDA Quantum enables researchers and developers to write and deploy hybrid code, thus making optimal use of CPUs, GPUs, and QPUs. The company announced a new group of partners integrating CUDA Quantum into their platforms, including quantum hardware companies Anyon Systems, Atom Computing, IonQ, ORCA Computing, Oxford Quantum Circuits, and QuEra; quantum software companies Agnostiq and QMware; and supercomputing centers National Institute of Advanced Industrial Science and Technology, the IT Center for Science (CSC), and the National Center for Supercomputing Applications (NCSA).