This is a Dockerfile to use NVIDIA SDK Manager on Docker container.
Ubuntu: 20.04
Nvidia SDKManager: 2.0.0
NVIDIA released official Docker image(https://docs.nvidia.com/sdk-manager/docker-containers/index.html).
- Docker
Please download the package of NVIDIA SDK Manager from https://developer.nvidia.com/nvidia-sdk-manager.
And, please put the package of NVIDIA SDK Manager in the same directory as the Dockerfile.
This time, I used sdkmanager_2.0.0-11405_amd64.deb.
docker build --build-arg GID=$(id -g) --build-arg UID=$(id -u) -t jetpack .
To build a Docker image with a specific SDK Manager version override the SDK_MANAGER_VERSION variable in the Docker command line
docker build --build-arg SDK_MANAGER_VERSION=2.0.0-11405 --build-arg GID=$(id -g) --build-arg UID=$(id -u) -t jetpack .
./launch_container.sh
Please launch NVIDIA SDK Manager by the following command.
sdkmanager
or for CLI version in terminal. I could not get the web browser to load to log in.
sdkmanager --cli
Download and attempt to install everything. Install fails, but all the files are downloaded. Stop SDKManger and run these commands
cd /home/jetpack/nvidia/nvidia_sdk/JetPack_5.1.3_Linux_JETSON_AGX_XAVIER_TARGETS/Linux_for_Tegra
sudo ./apply_binaries.sh
sudo ./tools/l4t_flash_prerequisites.sh
sudo ./flash.sh jetson-agx-xavier-devkit nvme0n1p1Please refer to https://docs.nvidia.com/sdk-manager/install-with-sdkm-jetson/index.html. And, I tested in the following setting.
- Manual Setup
- OEM Configuration: Runtime
If you get errors like this, it means that QEMU is not installed in host/container
To configure QEMU, run below, either on host, or the running container.
./configure_qemu.sh
