NVIDIA has just announced three new quantum computing developments at its recent GTC GPU Developers Conference.

We reported last year on a NVIDIA software tool called cuQuantum that was an SDK for quantum simulation on GPUs. By utilizing the Tensor core GPUs, it can significantly speed up simulations with the flexibility of using either the state vector or the tensor network methods of simulation depending upon which would be best for a particular algorithm. NVIDIA has been beta testing this tool and just announced that it has reached general release.

Their next announcement was for the DGX Quantum Appliance which is a full software stack that can run multi-GPU quantum circuit simulations using the cuQuantum libraries. This appliance works with the NVIDIA DGX A100 which is a very high performance hardware box that can provide up to 5 petaFlops of AI performance with up to eight of the NVIDIA A100 80GB Tensor Core GPUs in a 6U form factor. This product is currently using Google’s CIRQ software as its front end and is now in beta test.

The last announcement is the development of a quantum compiler that they called nvq++ for hybrid classical/quantum computing. It will use Quantum Intermediate Representation (QIR), a specification of a low-level machine language that hybrid computation that is based upon the LLVM intermediate language standard used in the classical computing. (See our article describing an alliance formed to support QIR here.) The nvq++ will initially work with the C++ programming language with added extensions for quantum and is targeted to support a wide range of backends, including both quantum processors and quantum simulators. This product is still in development with anticipated release later this year.

In addition, NVIDIA has been busy partnering with both several quantum hardware and software providers to provide mutual support for each other’s products. Organizations that NVIDIA has publicly announced they are working with include IBM, Google, IonQ, Xanadu, QC Ware, Classiq, and Oak Ridge National Laboratory. For more about NVIDIA’s latest announcement, you can access a blog with these announcements here and quantum related videos from NVIDIA’s GTC conference, including a keynote from CEO Jensen Huang here. A press release from Xanadu describing their work with NVIDIA can be accessed here. Another one from Classiq is available here.

March 22, 2022