A groundbreaking demonstration by Quantum Machines has shown an end-to-end NVIDIA CUDA-Q program running on live qubits combined with a classical PPU processor using NVIDIA NVQLink, marking a significant milestone in hybrid quantum-classical application development. This innovation merges Quantum Machines’ quantum control capabilities with NVIDIA’s CUDA-Q platform for quantum-GPU computing, alongside the NVQLink architecture, which facilitates a high-speed connection between quantum controllers and accelerated computing systems. This integration enables developers to craft quantum applications using widely recognized programming languages such as Python, C++, or QUA, bypassing the need for manually creating low-level control sequences typically essential for quantum hardware.
During the demonstration, code written with CUDA-Q was executed using Quantum Machines’ control stack across a quantum processor, GPUs, and CPUs, effectively routing various parts of workloads to the suitable processor. NVIDIA NVQLink plays a crucial role by enabling swift communication between the quantum processor and classical computing resources, with the entire process completed in roughly one microsecond. This technology is being showcased at the IEEE Quantum Week in Toronto, offering researchers and engineers a chance to observe its operation with live quantum hardware. Yonatan Cohen, CTO of Quantum Machines, expressed satisfaction with the collaboration with NVIDIA, emphasizing how these tools empower quantum developers to progress toward large-scale quantum computing more rapidly.
Traditionally, quantum processors, or QPUs, require specialized programming and control expertise. However, the latest integration aims to simplify this by making QPUs function like another computing resource within a larger system, working in tandem with CPUs and GPUs. Sam Stanwyck, Director of Quantum Product at NVIDIA, highlighted the transformative potential of quantum processors when tightly integrated with GPUs and CPUs, forming a unified quantum supercomputing system. Quantum Machines has incorporated NVIDIA NVQLink into its Quantum Machines Orchestration Platform, linking the hardware that controls and reads qubits with NVIDIA’s accelerated computing via a low-latency connection.
The ability to execute quantum operations on a QPU while utilizing CPUs and GPUs for classical processing in real-time represents a significant advancement. Quantum Machines’ control system converts these operations into precisely timed signals necessary for controlling and measuring qubits. This low-latency connection is particularly vital for workloads demanding rapid interaction between quantum and classical processors. It allows measurement data to be sent to classical processors and processing decisions to be returned to the quantum control system within microseconds.
This capability is poised to support future applications requiring real-time quantum-classical coordination, such as quantum error correction and other advanced quantum computing workloads. Quantum Machines and NVIDIA continue to refine low-latency connections between quantum processors and accelerated computing systems, with the latest demonstration focusing on advancing more accessible and scalable quantum computing.
Legal Disclaimer:
The information contained in this article has been provided by independent third-party contributors, clients, or content partners. We do not independently verify the accuracy, completeness, legality, ownership, licensing, or reliability of submitted content, including text, images, videos, trademarks, or other media materials. The submitting party is solely responsible for ensuring that all content, including images and media assets, complies with applicable copyright, trademark, licensing, and intellectual property laws. We disclaim liability for any unauthorized use of copyrighted or proprietary materials by third parties. If you believe that any content published on this platform infringes your intellectual property rights, kindly contact the author above for prompt review and resolution.