JetPack 5.1.1 is now live

[suhash]

Pleased to announce that JetPack 5.1.1 is now live.

JetPack 5.1.1 is the latest production release and brings support to our newly announced Jetson Orin Nano Developer Kit. It also adds support for Jetson AGX Orin 64GB, Jetson Orin NX 8GB, Jetson Orin Nano 8GB and Jetson Orin Nano 4GB production modules. With these additions JetPack 5.1.1 supports all Jetson Orin and Jetson Xavier modules and developer kits.

JetPack 5.1.1 includes Jetson Linux 35.3.1 with following highlights:

(Please read Jetson Linux Release notes for more details)

• Adds support for Jetson Orin Nano Developer Kit
• Adds support for Jetson AGX Orin 64GB, Jetson Orin NX 8GB, Jetson Orin Nano 8GB and Jetson Orin Nano 4GB production modules
• Over The Air Updates:
  • Image Based OTA tools to upgrade Xavier or Orin based modules running JetPack 5 in the field1
• Camera:
  • Support for Multi Point Lens Shading Correction (LSC) on Orin.​
  • Enhanced resilience of Argus SyncStereo app to maintain synchronization between stereo camera pairs
• Multimedia
  • Support for dynamic frame rate in AV1 encoding
  • New argus_camera_sw_encode sample for demonstrating software encoding on CPU cores
  • Updated nvgstcapture-1.0 with option of software encoding on CPU cores

Installing JetPack 5.1.1

You can install JetPack 5.1.1 using multiple ways:

1. Using SDK Manager: Install JetPack 5.1.1 using SDK Manager on any Jetson Orin and Xavier modules.
2. SDCard Image: If you are using Jetson Orin Nano Developer Kit or Jetson Xavier NX Developer kit, then you can use the SDCard image from JetPack 5.1.1 page.
3. Flashing Scripts: Download the Jetson Linux tar balls from Jetson Linux 35.3.1 page and use flashing scripts to flash. Refer to QuickStart section in Jetson Linux Developer Guide.
4. Debian Packages: After flashing Jetson Linux, you can install the rest of the JetPack using apt commands. Please refer JetPack documentation

To upgrade from a previous version of JetPack 5 using debian packages, refer to JetPack documentation.

JetPack 5.1.1 Components

• Jetson Linux 35.3.1
• CUDA 11.4.19
• TensorRT 8.5.2
• cuDNN 8.6.0
• VPI 2.2
• OpenCV 4.5.4
• Vulkan 1.3
• Nsight Systems 2022.5
• Nsight Graphics 2022.6
• Nsight DLD/Compute 2022.2

Resources

• JetPack 5.1.1 Documentation
• Jetson Linux 35.3.1 Developer Guide
• Jetson Linux 35.3.1 Release Notes

Important Note

Following items are pending to be released (we will update this announcement when they are made live):

• Jetson containers for JetPack 5.1.1 on NGC

[john.ojanen]

I tried install Jetpack 5.1.1 on a Jetson Xavier NX running Jetpack 5.1 from a m.2 SSD using the instructions for upgrading using the instructions from How to Install JetPack :: NVIDIA JetPack Documentation. It fails when attempting to install nvidia-l4t-kernel. After rebooting the ethernet port is not working and not listed with lspci. Luckily wifi still works.

cat /etc/nv_tegra_release yields

R35 (release), REVISION: 3.1, GCID: 32827747, BOARD: t186ref, EABI: aarch64, DATE: Sun Mar 19 15:19:21 UTC 2023

This is the error received:

Reading package lists… Done
Building dependency tree
Reading state information… Done
Calculating upgrade… Done
Get more security updates through Ubuntu Pro with ‘esm-apps’ enabled:
libavformat58 python2.7-dev libavfilter7 libpython3.9-minimal libswresample3
libpython3.9-stdlib libzmq5 python2.7-minimal libpython2.7 python2.7
python3.9 libpython2.7-dev libpostproc55 python3.9-minimal libavcodec58
libjs-jquery-ui libavutil56 libswscale5 libopenexr24 libsdl2-2.0-0
libmysofa1 libpython2.7-minimal libpython2.7-stdlib
Learn more about Ubuntu Pro at Ubuntu Pro | Ubuntu
0 upgraded, 0 newly installed, 0 to remove and 0 not upgraded.
5 not fully installed or removed.
After this operation, 0 B of additional disk space will be used.
Do you want to continue? [Y/n] y
Setting up nvidia-l4t-kernel (5.10.104-tegra-35.3.1-20230319081403) …
Using the existing boot entry ‘primary’
3668-100—1–jetson-xavier-nx-devkit-qspi-
Info. Installing mtdblock.
Info. Active boot storage: nvme0n1
Info. Legacy mode: false
TNSPEC 3668-300-0000-B.0-1-2-jetson-xavier-nx-devkit-qspi-
COMPATIBLE_SPEC 3668-100—1–jetson-xavier-nx-devkit-qspi-
TEGRA_LEGACY_UPDATE false
TEGRA_BOOT_STORAGE nvme0n1
TEGRA_EMMC_ONLY false
TEGRA_CHIPID 0x19
TEGRA_OTA_BOOT_DEVICE /dev/mtdblock0
TEGRA_OTA_GPT_DEVICE /dev/mtdblock0
Info: Write TegraPlatformCompatSpec with 3668-100—1–jetson-xavier-nx-devkit-qspi-.
Starting kernel post-install procedure.
Rootfs AB is not enabled.
ERROR. Procedure for kernel update FAILED.
Cannot install package. Exiting…
dpkg: error processing package nvidia-l4t-kernel (–configure):
installed nvidia-l4t-kernel package post-installation script subprocess returned error exit status 1
dpkg: dependency problems prevent configuration of nvidia-l4t-kernel-headers:
nvidia-l4t-kernel-headers depends on nvidia-l4t-kernel (= 5.10.104-tegra-35.3.1-20230319081403); however:
Package nvidia-l4t-kernel is not configured yet.

dpkg: error processing package nvidia-l4t-kernel-headers (–configure):
dependency problems - leaving unconfigured
No apport report written because the error message indicates its a followup error from a previous failure.
dpkg: dependency problems prevent configura
tion of nvidia-l4t-jetson-io:
nvidia-l4t-jetson-io depends on nvidia-l4t-kernel (>> 5.10.104-tegra-35.3-0); however:
Package nvidia-l4t-kernel is not configured yet.
nvidia-l4t-jetson-io depends on nvidia-l4t-kernel (<< 5.10.104-tegra-35.4-0); however:
Package nvidia-l4t-kernel is not configured yet.

dpkg: error processing package nvidia-l4t-jetson-io (–configure):
dependency problems - leaving unconfigured
dpkg: dependency problems prevent configuration of nvidia-l4t-display-kernel:
nvidia-l4t-display-kernel depends on nvidia-l4t-kernel (= 5.10.104-tegra-35.3.1-20230319081403); however:
Package nvidia-l4t-kernel is not configured yet.

No apport report written because the error message indicates its a followup error from a previous failure.
No apport report written because MaxReports
is reached already
dpkg: error processing package nvidia-l4t-display-kernel (–configure):
dependency problems - leaving unconfigured
dpkg: dependency problems prevent configuration of nvidia-l4t-kernel-dtbs:
nvidia-l4t-kernel-dtbs depends on nvidia-l4t-kernel (= 5.10.104-tegra-35.3.1-20230319081403); however:
Package nvidia-l4t-kernel is not configured yet.

dpkg: error processing package nvidia-l4t-kernel-dtbs (–configure):
dependency problems - leaving unconfigured
No apport report written because MaxReports is reached already
Errors were encountered while processing:
nvidia-l4t-kernel
nvidia-l4t-kernel-headers
nvidia-l4t-jetson-io
nvidia-l4t-display-kernel
nvidia-l4t-kernel-dtbs
E: Sub-process /usr/bin/dpkg returned an error code (1)

As this was already running Jetpack 5.1 this seemed to be the correct way to update. Was it? Is there any way to fix the current state or do I need to completely reinstall with the SDK Manager?

[john.ojanen]

For those curious, I ended up flashing with the SDKManager and it worked fine. Unfortunately the containers for L4T 35.3 are not available so its back to Jetpack 5.1 for a while yet.

[suhash]

@john.ojanen the containers from Jetpack 5.1 is known to work on JetPack 5.1.1. You can use them in the meantime

[john.ojanen]

I did try using the l4t-tensorflow (l4t-tensorflow:r35.2.1-tf2.11-py3) container but ran into problems when using pytorch/torchvision, which are installed using Nvidia’s instructions. I think the errors are due to the different versions of CUDA in 5.1 vs 5.1.1. I did try Pytorch 2.0.0 and both torchvision 0.14.1 and 0.15.1 but they all produced errors. I’ll try this again when I get the time.

[dusty_nv]

@john.ojanen

but ran into problems when using pytorch/torchvision, which are installed using Nvidia’s instructions.

Hi @john.ojanen, was this using the l4t-pytorch container, or installing the PyTorch wheel outside of container? Feel free to open a new topic about installing PyTorch/torchvision if you’re having problems with it - I recommend following the install instructions from this topic: PyTorch for Jetson. The version of CUDA is the same between JetPack 5.1 and JetPack 5.1.1 (CUDA 11.4.315)

[yilinw]

Hi everyone,

I am using the Nvidia SDK manager to set up my Jetson Xavier NX developer kit. The SDK manager is on the latest version 1.9.2.10888. I tried to follow the instruction several time but it fails.

My jetson is flashed and it is okay to run. However, I guess it is not being fully setup since after I open the terminal and type docker and nvidia-smi, it says command not found.

The log from the host terminal is listed below:
1. CUDA Runtime
15:55:55 ERROR: CUDA Runtime - target: E: Unable to locate package libfreeimage-dev

15:55:55 ERROR: CUDA Runtime - target: E: Unable to locate package libopenmpi-dev

15:55:55 ERROR: CUDA Runtime - target: E: Package ‘openmpi-bin’ has no installation candidate

15:55:55 ERROR: CUDA Runtime - target: Error: safe apt install failed: sudo apt-get install -y gnupg libgomp1 libfreeimage-dev libopenmpi-dev openmpi-bin, exit code: 100, retry number: 2

15:55:55 INFO: CUDA Runtime - target: will sleep 20 seconds before next attempt

15:56:15 INFO: CUDA Runtime - target: Reading package lists… 0% Reading package lists… 100% Reading package lists… Done

15:56:15 INFO: CUDA Runtime - target: Building dependency tree… 0% Building dependency tree… 0% Building dependency tree… 50% Building dependency tree… 50% Building dependency tree

15:56:15 INFO: CUDA Runtime - target: Reading state information… 0% Reading state information… 0% Reading state information… Done

15:56:15 INFO: CUDA Runtime - target: Package openmpi-bin is not available, but is referred to by another package.

15:56:15 INFO: CUDA Runtime - target: This may mean that the package is missing, has been obsoleted, or

15:56:15 INFO: CUDA Runtime - target: is only available from another source

15:56:15 ERROR: CUDA Runtime - target: E: Unable to locate package libfreeimage-dev

15:56:15 ERROR: CUDA Runtime - target: E: Unable to locate package libopenmpi-dev

15:56:15 ERROR: CUDA Runtime - target: E: Package ‘openmpi-bin’ has no installation candidate

15:56:15 ERROR: CUDA Runtime - target: Unhandled error when running sudo apt-get install -y gnupg libgomp1 libfreeimage-dev libopenmpi-dev openmpi-bin: E: Unable to locate package libfreeimage-dev;E: Unable to locate package libopenmpi-dev;E: Package ‘openmpi-bin’ has no installation candidate;

15:56:15 INFO: CUDA Runtime - target: [ Package PostInstall Finished with Error ]

15:56:15 INFO: CUDA Runtime - target: [ NV_L4T_CUDA_RUNTIME_TARGET_POST_INSTALL_COMP Install took 7m54s ]

15:56:15 ERROR: CUDA Runtime - target: command terminated with error

15:56:15 SUMMARY: CUDA Runtime - target: First Error: safe_apt_install failed to run.

2. CUDA Toolkit for L4T:

15:11:46 INFO: CUDA Toolkit for L4T - target: verifying checksum of /data/ssd/NvidiaJetsonXavierNX/cuda-repo-l4t-11-4-local_11.4.19-1_arm64.deb

15:11:52 INFO: CUDA Toolkit for L4T - target: Found file /data/ssd/NvidiaJetsonXavierNX/cuda-repo-l4t-11-4-local_11.4.19-1_arm64.deb with correct checksum, skip downloading.

15:12:05 INFO: CUDA Toolkit for L4T - target: verifying checksum of /data/ssd/NvidiaJetsonXavierNX/cuda-repo-l4t-11-4-local_11.4.19-1_arm64.deb

15:12:08 INFO: CUDA Toolkit for L4T - target: Found file /data/ssd/NvidiaJetsonXavierNX/cuda-repo-l4t-11-4-local_11.4.19-1_arm64.deb with correct checksum, skip downloading.

15:56:15 SUMMARY: CUDA Toolkit for L4T - target: Depends on failed component

My Jetson device is able to start up and running, but in the terminal I typed docker and nvidia-smi. It says command not found. Then, I gues it is not well-setup.

I was using this SDK manager to set up the Jetson successfully but not this time.

Sincerely,
Yilin Wang

[yilinw]

Please let me know if any additional information would help.

[igalNV]

Hi Yilin,
It seems to be failing to get a package from a public repository. Please validate the connection to the internet before re-try the install.
You can try install the packages directly by executing the command at the Jetson device: apt-get install -y gnupg libgomp1 libfreeimage-dev libopenmpi-dev openmpi-bin
Also, if you can provide the logs by clicking “Export Logs” at STEP 04 or grab it from ~/.nvsdkm/ folder.

Ilan.

[john.ojanen]

I was installing PyTorch from the instructions. I’m fairly sure the problem I had with torchvision was because the container runtime wasn’t set to nvidia. I’d try Jetpack 5.1.1 again if but updating with the package manager from the instructions here (How to Install JetPack :: NVIDIA JetPack Documentation) did not work and I’m wary of trying again. Using the SDK Manager only seems to work when the device is in recovery mode.

[dusty_nv]

@john.ojanen

Using the SDK Manager only seems to work when the device is in recovery mode.

For SDK Manager to flash/reflash the base OS image over micro USB, yes the device needs to be in recovery mode. In order for it to install packages like CUDA/cuDNN/TensorRT/ect, it just needs to be connected to the same network as your PC. You can install those same packages from apt however without needing SDK Manager or your PC.

[yilinw]

Hi,

I checked the internet. I have the internet on the host, and I also set up the wifi for the jetson. Previously, I don’t need to do this step, but I did this time, and it still have the same failure. I guess the Jetson is using usb0 and rndis0 to talk to my host machine SDK manager.

I also try to manually install those packages, but libfreeimage-dev and libopenmpi-dev openmpi-bin are unable to be located and no installation candidate.

The logs from the folder is below
sdkm-2023-04-18-11-40-06.log (1.1 MB)

Sincerely,
Yilin Wang

[EdwardZhou]

Hi yilinw,

In the log, I could see the following error:
Err:6 Index of /ubuntu-ports focal-security InRelease
12:40:18.313 - info: NV_L4T_OPENCV_RUNTIME_TARGET_POST_INSTALL_COMP@JETSON_XAVIER_NX_TARGETS: Temporary failure resolving ‘ports.ubuntu.com’

This error seems to be caused by temporary ISP DNS issue. That’s why the issue also exists when you manually install the packages but also fail.
Can you ping ports.ubuntu.com successfully?

[yilinw]

Hi,

Thank you for the reply. I guess I can ping ports.ubuntu.com.
yilin@rickslade:~$ ping ports.ubuntu.com
PING ports.ubuntu.com (185.125.190.39) 56(84) bytes of data.
64 bytes from aerodent.canonical.com (185.125.190.39): icmp_seq=1 ttl=53 time=93.1 ms
64 bytes from aerodent.canonical.com (185.125.190.39): icmp_seq=2 ttl=53 time=93.5 ms

Please let me know if there is anything inappropariate. Thank you very much.

Sincerely,
Yilin Wang

[EdwardZhou]

Hi yilinw,
Since now the temporary DNS issue is solved, can you please retry installation with SDK Manager and see if the installation is fine?

[1354065425]

Unable to install onnxsim on Jetson xavier nx