cuQuantum Appliance

Type: Platform Tags: NVIDIA, cuQuantum, NGC, container, Qiskit, Cirq, qsim, quantum simulation, multi-GPU Related: cuQuantum, cuStateVec, cuTensorNet, cuDensityMat, cuPauliProp, cuStabilizer, CUDA-Q, NVIDIA-Quantum, NGC, NVIDIA-Container-Toolkit, NVIDIA-GPU-Operator, NCCL Sources: https://docs.nvidia.com/cuda/cuquantum/latest/appliance/overview.html, https://docs.nvidia.com/cuda/cuquantum/latest/ Last Updated: 2026-04-29

Summary

The NVIDIA cuQuantum Appliance is a containerized multi-GPU, multi-node quantum circuit simulation solution. It packages the current cuQuantum libraries, provides optimized frontends for Qiskit Aer through cusvaer and Google Cirq through qsim, and distributes examples and runtime dependencies through an NGC container image.

Detail

Purpose

Quantum researchers often want GPU-accelerated simulation without assembling every CUDA, Python, MPI, framework, and quantum-library dependency by hand. cuQuantum Appliance gives them a supported container path for running Qiskit and Cirq workflows on NVIDIA GPUs while using cuQuantum libraries underneath.

Current scope

  • Containerized access to cuStateVec, cuTensorNet, cuDensityMat, cuPauliProp, and cuStabilizer.
  • Distributed state-vector backend for IBM Qiskit Aer through cusvaer.
  • Optimized multi-GPU Google Cirq frontend through qsim.
  • NGC container image flow with Docker, NVIDIA GPU drivers, and NVIDIA-Container-Toolkit as host prerequisites.
  • Example workloads such as GHZ, hidden-shift, and Simon examples inside the container.
  • x86_64 and arm64 image variants documented through the ${march} tag pattern.

NVIDIA context

cuQuantum Appliance is the deployment/distribution surface for cuQuantum simulation workflows. It belongs near NGC, NVIDIA-Container-Toolkit, and NVIDIA-GPU-Operator for operational context, while the component libraries remain separate pages because they have distinct APIs and workloads.

Connections

Resources