NVIDIA and Quantum Machines have introduced a new system called DGX Quantum that combines technologies from both companies for providing programming and control of a quantum processor. The Quantum Machines portion is the OPX+ Quantum Control Platform which provide the waveforms for controlling the qubits with the low level waveforms and measurement to operate the qubits. The NVIDIA portion will include a new Grace Hopper Superchip that couples a very high performance GPU (Hopper) with a very high performance CPU (Grace) on the same module. The system is expandable and multiple Grace Hopper Superchips can be connected to the OPX+ through the PCI bus. This module is connected by a very high speed PCI bus to the Quantum Machines OPX+ that can provide real time latencies of a few hundred nanoseconds. The system can be controlled with NVIDIA’s CUDA Quantum programming model, formerly called QODA which was introduced last year. NVIDIA has also made the CUDA Quantum platform open source.
The programming language used within CUDA Quantum is called NVQ++ which provides extensions for both C++ and Python to support hybrid classical-quantum systems. The compiler toolchain can produce intermediate level code that is compatible with Multi-Level Intermediate Representation (MLIR) and Quantum Intermediate Representation (QIR). This will allow them to support different types of quantum processors. The combination of the higher performance NVIDIA and Quantum Machines units can support uses such as real-time error correction and mid-circuit measurements. These functions require very short latencies because they need to be accomplished in a time much shorter than the coherence time of the qubit.
NVIDIA has announced several new partners that are integrating CUDA Quantum into their platforms including 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). The is in additional to previous partners they have announced including IQM, Pasqal, Quantinuum, Quantum Brilliance, Xanadu, QMWare, QC Ware, and Zapata. One of the early customers for the DGX Quantum product will be the Israel Quantum Computing Center. For smaller companies that may not have all the resources to develop a full-stack, high performance hybrid classical/quantum architecture in-house, this solution allows them to bring in this capability and be more competitive with some of the larger companies that are offering this.
Additional information about this announcement can be found in a press release here, a blog post here, a video excerpt about the announcement from a keynote address at GTC 2023 by NVIDIA CEO Jensen Huang here, an associated press release from QMWare here, and another associated press release from Quantum Brilliance here.
March 21, 2023