Is this page helpful?

System Requirements#

Windows, Linux, Windows Subsystem for Linux (WSL), Jetson Thor, and DGX Spark#

PyNvVideoCodec is supported on Windows, Linux, Windows Subsystem for Linux (WSL), NVIDIA Jetson Thor platforms, and NVIDIA DGX Spark. The following requirements apply to all supported platforms:

Table 1 System Requirements#

Operating System

Windows 10 or higher (x64); Ubuntu 20.04 or higher (Linux x86_64 and aarch64); Ubuntu 20.04 or higher for WSL; Jetson Linux 39.2 or higher (Jetson Thor, Linux arm64); NVIDIA DGX OS (DGX Spark, Linux arm64)

GPU

Discrete GPU platforms: Turing, Ampere, Ada, Hopper, Blackwell

Integrated platform GPUs: NVIDIA Jetson Thor, NVIDIA DGX Spark

Drivers

NVIDIA Windows display driver 531.61 or newer; Get most recent NVIDIA Display Driver

Jetson Thor: Jetson Linux 39.2 or higher

DGX Spark: NVIDIA DGX OS with the NVIDIA driver and CUDA stack supplied by the NVIDIA DGX Spark system image

Python

Windows x64 wheels: Python 3.10 to 3.14

Linux x86_64 wheels: Python 3.10 to 3.14, built on a manylinux base image with glibc 2.28 and GCC 11.x; requires glibc 2.28 or newer at runtime

Linux arm64/aarch64 wheels: Python 3.10–3.12 (glibc 2.28, GCC 11.x; requires glibc 2.28 or newer at runtime); Python 3.13–3.14 (glibc 2.34; requires glibc 2.34 or newer at runtime)

Building from source on Ubuntu: Python development headers for the selected version are required, for example python3.10-dev, python3.11-dev, python3.12-dev, python3.13-dev, or python3.14-dev.

CMake

3.21 and onwards

Visual Studio (Windows only)

Visual Studio

CUDA Toolkit

Discrete GPU platforms: Latest CUDA Toolkit; Jetson Thor and DGX Spark: CUDA 13.0 or higher from the platform software stack

Python modules to run Sample applications

PyCUDA and PyTorch

Additional Configuration for WSL#

In addition to all the requirements listed above, WSL users need the following configuration:

  • Add the directory /usr/lib/wsl/lib to the PATH environment variable if it is not added by default. This is required to ensure the WSL libraries are included in the system path.

Additional Configuration for Jetson Thor#

In addition to all the common Linux requirements listed above, Jetson Thor users need the following configuration:

  • Use a Jetson Linux release that is compatible with the PyNvVideoCodec package and the NVIDIA Video Codec SDK release used by the package.

  • Install CUDA from the Jetson Linux or JetPack package repositories. If CUDA headers and development libraries are required, install the Jetson platform CUDA development packages, such as nvidia-cuda-dev.

  • Use Linux arm64 Python wheels and Python dependencies that match the Jetson Linux CUDA stack.

Additional Configuration for DGX Spark#

In addition to all the common Linux requirements listed above, DGX Spark users need the following configuration:

  • Use NVIDIA DGX OS and keep the system image updated to receive the recommended NVIDIA driver, CUDA, and library stack for DGX Spark.

  • Use Linux arm64 Python wheels and Python dependencies that match the DGX OS CUDA stack.