How NVIDIA CUDA-Q Unites CPUs, GPUs, and QPUs
With the release of the open-source NVIDIA CUDA-Q platform, NVIDIA declared that it will expedite quantum computing work at national supercomputing centres worldwide.
Poland, Japan, and Germany have supercomputers Advance Quantum Computing Research by Including Grace-Hopper and Quantum-Classical Accelerated Supercomputing Platform
NVIDIA CUDA-Q
The cutting-edge hybrid quantum-classical computer platform
Dynamic workflows spanning system architectures require a bridging technology for algorithm research and quantum advantage applications. NVIDIA CUDA-Q is an open-source platform that combines and programmes GPUs, CPUs, and quantum processing units (QPUs) in a single system using a unified and open programming model. GPU-accelerated system performance and scalability across heterogeneous QPU, CPU, GPU, and emulated quantum system elements are made possible by NVIDIA CUDA-Q.
In order to facilitate the creation of hybrid applications, NVIDIA CUDA-Q provides a single programming paradigm intended for a hybrid environment in which CPUs, GPUs, and QPUs collaborate. It comprises of a system-level toolchain that facilitates application acceleration and language extensions for Python and C++.
Principal Advantages
Productive
Simplifies the creation of hybrid quantum-classical systems using a single programming model, increasing the efficiency and scalability of research on quantum algorithms.
Adaptable Structure
Interfaces with contemporary GPU-accelerated apps, integrates easily with toolchains, and connects to partner QPUs and GPU simulators.
Excellent Outcomes
Up to 26 qubits can be simulated with a 2500X speedup on four A100 GPUs, and 40 qubits can be simulated by spreading the simulation across 128 GPU nodes.
The platform will power the quantum processing units (QPUs) within NVIDIA-accelerated high-performance computing systems at supercomputing centres in Germany, Japan, and Poland.
Quantum processor units (QPUs) are the brains of quantum computers. They can potentially perform some computations more quickly than conventional processors by using the behaviour of particles like electrons or photons in their calculations.
IQM Quantum Computers’ QPU will supplement Jülich Supercomputing Centre (JSC) at Forschungszentrum Jülich’s JUPITER supercomputer, powered by the NVIDIA GH200 Grace Hopper Superchip.
Japan’s National Institute of Advanced Industrial Science and Technology (AIST) is home to the ABCI-Q supercomputer, which is intended to further the country’s quantum computing project. Equipped with a QuEra QPU, the machine will be powered by the NVIDIA Hopper design.
The Poznan Supercomputing and Networking Centre (PSNC) in Poland has integrated a pair of photonic QPUs manufactured by ORCA Computing. These QPUs are linked to a freshly established supercomputer partition that is powered by NVIDIA Hopper.
Tim Costa, NVIDIA’s head of quantum and HPC, stated that “tight integration of quantum with GPU supercomputing will enable useful quantum computing.” Pioneers like AIST, JSC, and PSNC are able to push the limits of scientific discovery and improve the state of the art in quantum-integrated supercomputing thanks to NVIDIA’s quantum computing platform.
By using laser-controlled Rubidium atoms as qubits to execute calculations, researchers at AIST will be able to explore quantum applications in AI, energy, and biology thanks to the integration of QPU with ABCI-Q. These atoms are identical to those found in precision atomic clocks. Since every atom is the same, this offers a promising way to develop a high-fidelity, large-scale quantum processor.
Masahiro Horibe, deputy director of G-QuAT/AIST, stated that “Japan’s researchers will make progress towards practical quantum computing applications with the ABCI-Q quantum-classical accelerated supercomputer.” “These innovators are pushing the limits of quantum computing research with NVIDIA’s assistance.”
With two PT-1 quantum photonics devices, PSNC’s QPUs will allow researchers to investigate biology, chemistry, and machine learning. Single photons, or light packets at telecom frequencies, are used by the systems as qubits. This makes it possible to use readily available, standard telecom components to create a distributed, scalable, and modular quantum architecture.
According to Krzysztof Kurowski, CTO and deputy director of PSNC, “our partnership with ORCA and NVIDIA has allowed us to create a unique environment and build a new quantum-classical hybrid system at PSNC.” Developers and users need open, easy deployment and programming of multiple QPUs and GPUs managed by user-centric services. A new generation of quantum-accelerated supercomputers for numerous cutting-edge application fields is made possible by this close partnership.
Researchers at JSC will be able to create quantum applications for chemical simulations and optimisation issues thanks to the QPU’s integration with JUPITER. They will also be able to show how quantum computers can speed classical supercomputers. Superconducting qubits, or electrical resonant circuits, which can be produced to function at low temperatures like artificial atoms, are the building blocks of this device.
Head of JSC’s quantum information processing section Kristel Michielsen stated, “Hybrid quantum-classical accelerated supercomputing is bringing quantum computing closer.” “JSC researchers will further the fields of quantum computing, chemistry, and material science through our continuous collaboration with NVIDIA.”
CUDA-Q allows quantum computing with AI to tackle issues like noisy qubits and create effective algorithms by tightly integrating quantum computers with supercomputers.
An open-source, GPU-independent quantum-classical accelerated supercomputing platform is called CUDA-Q. The majority of businesses using QPUs utilise it because it offers best-in-class performance.
