JetPack 6.2 on Jetson Orin Nano with UTM on M3 Mac

I needed to set up my Jetson Orin Nano from scratch with a fresh OS.

Unlike other Jetson models with an SD card slot, mine uses an SSD, so I had to handle it differently.

To do this, I emulated Ubuntu 22.04 LTS (AMD64) on my MacBook M3 (ARM) using UTM to run NVIDIA's SDK Manager, which only supports AMD Linux. After setting everything up, I successfully installed a fresh OS on my Jetson Orin Nano from scratch.

Specs

JetPack 6.2
MacBook M3 (ARM)
Jetson Orin Nano
Ubuntu 22.04 LTS (AMD64)
- Note: AMD version, not ARM (ARM64/AArch64), as NVIDIA SDK Manager requires AMD.
Link

1. Install UTM

https://mac.getutm.app/

2. Install Ubuntu 22.04 LTS (Desktop)

https://releases.ubuntu.com/jammy/

Download ubuntu-22.04.5-desktop-amd64.iso

3. Run Ubuntu

Select "Emulate" to run AMD architecture on Mac Silicon (M3).

Config:
- Cores: 4
- RAM: 8192 MB

4. Install Ubuntu

Download

Note: Installation takes a while due to emulation.

After installation, eject the ISO and reboot.

5. Download NVIDIA SDK Manager

$ sudo apt install -y ./sdkmanager_2.2.0-12028_amd64.deb

6. Run SDK Manager

$ sdkmanager

Ignore the warning and proceed.

Select "Orin Nano".

7. Problem

The NVIDIA SDK Manager didn't recognize the Jetson Orin Nano, and neither did UTM's Ubuntu VM. I realized I needed an Arduino jumper cable to connect it properly.

Source: [Jetson] Jetson Orin Nano SSD에 Jetpack 설치 (SDK Manager 사용)

8. Solution: Jumper Cable

With the jumper cable, my MacBook M3 detected the NVIDIA device:

Mac can catch Nvidia usb

❯ system_profiler SPUSBDataType | grep -A 10 "APX"
        APX:
          Product ID: 0x7523
          Vendor ID: 0x0955  (NVIDIA)
          Version: 4.01
          Speed: Up to 480 Mb/s
          Manufacturer: NVIDIA Corp.
          Location ID: 0x01100000 / 1
          Current Available (mA): 500
          Current Required (mA): 32
          Extra Operating Current (mA): 0

But UTM's Ubuntu still couldn't see it:

ubuntu@ubuntu-Standard-PC-Q35-ICH9-2009:~$ lsusb
Bus 002 Device 004: ID 0627:0001 Adomax Technology Co., Ltd QEMU USB Keyboard
Bus 002 Device 003: ID 0627:0001 Adomax Technology Co., Ltd QEMU USB Mouse
Bus 002 Device 002: ID 0627:0001 Adomax Technology Co., Ltd QEMU USB Tablet
Bus 002 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub
Bus 004 Device 001: ID 1d6b:0001 Linux Foundation 1.1 root hub
Bus 003 Device 001: ID 1d6b:0001 Linux Foundation 1.1 root hub
Bus 001 Device 001: ID 1d6b:0001 Linux Foundation 1.1 root hub
Bus 006 Device 001: ID 1d6b:0003 Linux Foundation 3.0 root hub
Bus 005 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub

I tried increasing UTM's shared device count:

But it didn't work.

9. Solution: UTM USB-C Port Fix

I connected my Jetson (NVIDIA APX) to the right USB-C port on my MacBook M3, but it didn't show up in UTM's Ubuntu VM with lsusb. Switching it to the left USB-C port fixed it—lsusb now showed 0955:7523 NVIDIA Corp. APX. Same USB 3.1 bus, different port. A strange but effective solution!

Now, let's start the installation.

10. Error

During installation with NVIDIA SDK Manager, I hit an error:

plymouthd assert failure: ply-event-loop.c:917

This assertion error seemed related to a mismatch between UTM's display settings and SDK Manager.

11. Solution: Plymouthd Assertion Error

To avoid the plymouthd assertion error, I switched to CLI mode:

$ sdkmanager --cli --action install --login-type devzone --product Jetson --target-os Linux --version 6.2 --show-all-versions --host --target JETSON_ORIN_NANO_TARGETS --select 'Jetson Linux' --select 'Jetson Linux image' --select 'Flash Jetson Linux' --select 'Jetson Runtime Components' --select 'Additional Setups' --select 'DateTime Target Setup' --select Gstreamer --select 'DLA Compiler' --select 'CUDA Runtime' --select 'CUDA Runtime' --select 'CUDA X-AI Runtime' --select 'CuDNN Runtime' --select 'TensorRT Runtime' --select 'Computer Vision Runtime' --select 'OpenCV Runtime' --select 'CuPVA Runtime' --select 'VPI Runtime' --select 'NVIDIA Container Runtime' --select 'NVIDIA Container Runtime with Docker integration (Beta)' --select Multimedia --select 'Multimedia API' --deselect 'Host SDK Components' --deselect CUDA --deselect 'CUDA on Host' --deselect 'CUDA Cross Compile Package on Host' --deselect NvSci --deselect NvSci --deselect 'Computer Vision' --deselect 'VPI on Host' --deselect 'Developer Tools' --deselect 'Nsight Systems' --deselect 'Nsight Graphics' --deselect 'Nsight Perf SDK' --deselect 'Jetson SDK Components' --deselect CUDA --deselect 'CUDA Toolkit for L4T' --deselect 'CUDA-X AI' --deselect 'CuDNN on Target' --deselect 'TensorRT on Target' --deselect 'Computer Vision' --deselect OpenCV --deselect 'VPI on Target' --deselect 'Developer Tools' --deselect 'Nsight Systems' --deselect 'Nsight Graphics' --deselect 'Jetson Platform Services' --deselect 'Jetson Platform Services' --deselect 'Jetson Platform Services' --flash --license accept

It worked perfectly! 👍👍

제민욱 @minwook