Установка NVIDIA cuOpt на GPU-сервер

★ 7.6 · content

cuopt-install is a Claude Code skill that guides you through installing NVIDIA cuOpt for use via the Python API, C API, or as a REST server. It covers selecting the correct CUDA suffix (cu12 or cu13) to match your driver, installing via pip, conda, or Docker, verifying the installation, and resolving common errors such as missing modules or CUDA version mismatches. One key detail the skill clarifies: installing the Python package `cuopt-cuXX` automatically pulls in `libcuopt-cuXX` as a dependency, so the C library and headers are immediately available without a separate step. Aimed at developers and MLOps engineers deploying cuOpt on GPU servers (Volta or newer, Compute Capability ≥ 7.0), cloud instances, or containerized environments.