Flashing the NVIDIA Jetson (JetPack 6.x)

Note: These steps target Jetson Orin NX/Nano on Connect Tech carriers (Hadron, etc.), Jetson Linux R36.3. Adjust versions as needed.

What you’ll do

  • Prepare a host machine and download JetPack BSP + sample rootfs
  • Install the CTI BSP on the host
  • Put the Jetson in recovery mode and flash (auto or manual)
  • Perform first boot steps and basic post-setup

Prerequisites

  • Host: Ubuntu 20.04 (x86_64)
  • USB cable for recovery/flash
  • NVMe M.2 SSD installed on the carrier

1) Get JetPack (Driver Package + RootFS)

1) Create a working directory, referred to as BSP_ROOT below.

2) Open the Jetson Linux R36.3 downloads and notes:

  • Driver page: https://developer.nvidia.com/embedded/jetson-linux-r363

3) Download both for Orin (t234):

  • Driver Package (BSP): Jetson_Linux_R36.3.0_aarch64.tbz2
  • Sample Root Filesystem: Tegra_Linux_Sample-Root-Filesystem_R36.3.0_aarch64.tbz2

4) Place both files under BSP_ROOT, then extract them:

cd <BSP_ROOT>
sudo tar -jxf Jetson_Linux_R36.3.0_aarch64.tbz2
sudo tar -C Linux_for_Tegra/rootfs/ -xjf Tegra_Linux_Sample-Root-Filesystem_R36.3.0_aarch64.tbz2

After extraction, you should have <BSP_ROOT>/Linux_for_Tegra populated.

2) Install the CTI BSP

If you used NVIDIA SDK Manager, BSP_ROOT is typically:

  • ~/nvidia/nvidia_sdk/<JetPack_Version>_Linux_JETSON_NX_ORIN_TARGETS/
  • or ~/nvidia/nvidia_sdk/<JetPack_Version>_Linux_JETSON_NANO_ORIN_TARGETS/

If you manually downloaded from the Embedded Download Center, use the folder you created as BSP_ROOT.

1) Copy the CTI package to Linux_for_Tegra:

cp CTI-L4T-ORIN-NX-NANO-36.3.0-V###.tgz <BSP_ROOT>/Linux_for_Tegra

2) Extract it and run the installer:

cd <BSP_ROOT>/Linux_for_Tegra
sudo tar -xzf CTI-L4T-ORIN-NX-NANO-36.3.0-V###.tgz
cd CTI-L4T
sudo ./install.sh
cd .. # back to Linux_for_Tegra

The CTI-L4T BSP is now installed on the host, and you can flash Orin NX/Nano.

3) Put the device in Recovery Mode

1) Connect the carrier to your host via USB per the carrier manual. 2) Ensure an NVMe M.2 SSD is installed. 3) Follow the carrier instructions to enter recovery (button/strap sequence).

Tip: You can verify via lsusb for an NVIDIA device in recovery.

4) Flashing options

You can use either the CTI automated script or the manual NVMe flash script.

Option A: Automated (menu-driven)

cd <BSP_ROOT>/Linux_for_Tegra
sudo ./cti-flash.sh

Follow the menu to select your module and board configuration. Flashing will begin.

Option B: Manual NVMe flash

Do not include the .conf extension in <config>.

cd <BSP_ROOT>/Linux_for_Tegra
sudo ./cti-nvme-flash.sh cti/<module>/<boardname>/<config>
# <module> is one of: orin-nx | orin-nano

Examples:

sudo ./cti-nvme-flash.sh cti/orin-nx/boson/base
sudo ./cti-nvme-flash.sh cti/orin-nano/boson/base

When flashing completes, the device will reboot.

5) First boot and basic setup

  • Guide: https://jetsonhacks.com/2019/04/19/jetson-nano-serial-console/
  • Complete the initial setup prompts. You can skip network here and do it below.

Join Wi‑Fi via nmcli

nmcli dev wifi rescan
nmcli d wifi connect <SSID> password <PASSWORD>

SSH in

ssh <user>@<device-ip>

6) Useful post-setup

NoMachine (remote desktop)

  • https://kb.nomachine.com/AR02R01074

Change power mode

sudo nvpmodel -m <mode-number>
  • Reference: https://forums.developer.nvidia.com/t/power-mode-in-terminal/287224

Install Docker + NVIDIA Container Toolkit + Git LFS

  • Docker: https://docs.docker.com/engine/install/ubuntu/
  • NVIDIA Container Toolkit: https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/latest/install-guide.html
  • Git LFS: https://git-lfs.com/

Build torchvision from source (matches NVIDIA PyTorch)

export FORCE_CUDA=1
git clone -b v0.17.2 https://github.com/pytorch/vision.git
cd vision
sudo -E python setup.py develop  # use 'install' if you don't need editable mode

Mount a USB drive

  • https://linuxconfig.org/howto-mount-usb-drive-in-linux

References

1) Hadron carrier: https://connecttech.com/product/hadron-carrier-for-nvidia-jetson-orin-nx/ 2) CTI L4T release notes (36.3.0): https://connecttech.com/ftp/Drivers/L4T-Release-Notes/Jetson-Orin-NX-Orin-Nano/ORIN-NX-NANO-36.3.0.pdf 3) CTI knowledge base: https://connecttech.com/resource-center/kdb373/ 4) torchvision dev install: https://github.com/pytorch/vision/blob/main/CONTRIBUTING.md#development-installation

Continue

← Back to Home View on GitHub