Nvidia continues to expand their offerings to the quantum community. Last year, they introduced their cuQuantum SDK for implementing high performance simulation of quantum programs using their Nvidia GPU’s. They have now announced a new platform called QODA (Quantum-Optimized Device Architecture) that also enables the development and compilation of quantum-classical programs for hybrid environments that include classical CPU’s, GPUs, and quantum QPU’s connected together. The basic programming interface is currently C++ with Python planned to be added in the future. And the compiler, which they call NVQ++, will utilize the QIR (Quantum Intermediate Representation) which will make it easier to add more hardware backends in the future.

Nvidia also announced multiple partnerships with organizations to support QODA that include hardware providers QM Quantum Computers, Pasqal, Quantinuum, Quantum Brilliance and Xanadu; software providers QC Ware and Zapata Computing; and supercomputing centers Forschungszentrum Jülich, Lawrence Berkeley National Laboratory and Oak Ridge National Laboratory and indicated they expect to add even more partners to this list in the coming months. The product is still in development and the availability for beta users is expected by the end of 2022 with general availability anticipated in early 2023.

Nvidia sees the introduction of QODA as a way to open up the access of quantum computing to a wider range of users. They see an analogy with their earlier introduction of a classical software platform in 2007 called CUDA (Compute Unified Device Architecture). At the time Nvidia’s GPU chips were mostly used as graphics accelerators and the only programming tools available were various graphics APIs. The introduction of CUDA open the way for subject matter experts in areas not related to graphics to utilize the high performance capabilities of GPUs in field completely unrelated to graphics. CUDA has been highly successful and has had over 30 million downloads since it was introduced.

There are other efforts working hard on facilitating the integration of quantum-classical programs. Most notably, IBM has announced their OpenQASM3 specification for programming along with their Qiskit Runtime operating software for optimizing the interfaces between the classical and quantum computers.

More information about QODA is available in a news release provided by Nvidia here, a QODA product page on their website here, and a video presentation here.

July 12, 2022