How to Install Jetson SDK Components (CUDA, cuDNN, etc) on NVIDIA Jetson Orin Series via SDK Manager

Now Jetson Orin Nano is ready to install further NVIDIA SDK components. Download the SDK Manager for host Ubuntu 20.04 or 22.04 and install it on host Ubuntu.

Connect the Jetson Device to the host machine using Ethernet in same LAN network connection of host Ubuntu. From a terminal window, launch SDK Manager with the command:

sdkmanager

Log in to your account, if the Jetson Orin Nano is recognized normally, SDK Manager will detect it correctly.

Check your Jetpack version. If your Jetpack version doesn't appear use the following command to launch the SDK manager with archived Jetpack versions.

sdkmanager --archived-version

Note: You can also check the Jetpack Version from the display of the Jetson Device. Just type down the following command in the display terminal.

sudo apt-cache show nvidia-jetpack

See Above Photo there is Target Hardware is Jetson Orin Nano Module Automatically Detected.

Select the Target Operating System to Linux Jetpack 5.1.5 and Click

Now you can proceed ahead by clicking on CONTINUE. Before that, you can tick on the DeepStram app if you wish to install it.

Next, uncheck Jetson OS (Because we have already flashed it using BSP), and check Jetson SDK Components(CUDA, Tensorrt, etc.). Check I accept the terms & conditions... and click CONTINUE.

 Then your display come above image Enter your sudo password of host PC then click OK

Then your display come above image Shown Click on Connection to Select option Ethernet and enter Your Jetson Orin Nano Ethernet IP Address.

Enter your Jetson Orin Nano Username and Password then click Install.

Note: Jetson Device (Orin Nano) to the host machine using Ethernet in same LAN network connection of host Ubuntu. 

Above image only for view Installation Process is Running.

Above image only for view Installation Process is Running.

Above image only for view Installation Process is Running.

Above image only for view Installation Process is Running.

After Complete Above image is comes.

If you need another Board installation then Click On Back to Step 01 otherwise Click On Finish and Exit Button.

Note: If you want to learn more about how to use NVIDIA SDK-manager, you can use the official guide from NVIDIA from here.

Once all the required NVIDIA SDK components are successfully installed you can proceed ahead to install PyTorch and torchvision from here.

Special thanks to our fellow Engineer and Colleague -Kishan & Dharmesh