Container Images for Video Codec SDK
This repo includes the source for building base container images for the NVIDIA Video Codec SDK.
Versions Supported
The build recipes include the following versions:
- Video Codec SDK: 11.0.10
- CUDA: 11.2.2
Getting Started
Build files are provided for the following Linux distributions:
- Ubuntu 18.04: (ubuntu18.04/Dockerfile)
Building Images
First, clone the repo:
$ git clone https://gitlab.com/nvidia/container-images/video-codec-sdk.git
Download the Video Codec SDK (11.0.10
) archive from the NVIDIA Video Codec SDK page and
place it in the image directory. Build the Docker container:
$ docker build --pull \
-t nvcr.io/nvidia/video-codec-sdk:11.0.10 \
--build-arg repository=nvcr.io/nvidia/cuda \
--build-arg flavor=devel \
--file Dockerfile .
Once the image is built, you can run the Docker container:
$ docker run --gpus all -it \
nvcr.io/nvidia/video-codec-sdk:11.0.10 \
bash
Running Samples
After launching the container in the previous step, you can build and run the samples from the container.
First, install the FFMpeg headers and libraries to allow the samples to build cleanly.
apt-get install -y \
libavutil-dev libavcodec-dev \
libavformat-dev libswresample-dev \
pkg-config
Now, we proceed to build the AppDec
sample and run it with a sample input video file.