JetPack 5.1 with Jetson Linux 35.2.1 Released

Robotics & Edge Computing Jetson & Embedded Systems Jetson Orin NX
1

We are very pleased to announce that JetPack 5.1 is now released!

JetPack 5.1 is a production release with added support for Jetson Orin NX 16GB production module. It also supports Jetson AGX Orin 32GB, Jetson AGX Orin Developer Kit, Jetson AGX Xavier series, Jetson Xavier NX series as well as Jetson AGX Xavier Developer Kit and Jetson Xavier NX Developer Kit.

JetPack 5.1 includes Jetson Linux 35.2.1 with new security and OTA features as well as enhancements for Multimedia, Camera and Display. Below are the top highlights for JetPack 5.1. Please refer to JetPack 5.1 page for full lists of features:

  • Adds support for Jetson Orin NX 16GB production module
  • Security
  • Over The Air Updates:
    • Image Based OTA tools with A/B support to upgrade Jetson AGX Xavier or Jetson Xavier NX modules running JetPack 4.x releases
    • Image Based OTA tools to upgrade Jetson Orin modules will be released with the next release
  • Display
    • Fixes for HDMI 2.0 and DP compliance
  • Camera
    • Support for AR1335 YUV camera
    • Enhanced support for simultaneous V4l2 and Argus
    • New Argus Demosaic (argus_demosaicOutput sample) to output RGB supported on Orin
    • New Argus RAW reprocessing (syncStereoRawReprocess sample) to read Bayer raw images from file & provide reprocessed YUV output to file
    • Updated Argus SyncSensorCalibrationData extension with enhanced EEPROM Calibration data to support Camera Module Serial Number, and IMU Noise Parameters
  • MultiMedia
    • TAQ( Temporal Adaptive Quantization) support is extended to H264 encoder

JetPack 5.1 Components:

  • Jetson Linux 35.2.1
  • 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

To upgrade from previous JetPack 5 Releases:

To upgrade from previous JetPack 5 releases, first edit /etc/apt/sources.list.d/nvidia-l4t-apt-source.list to point to the 35.2 repo (just change the version to r35.2 in both lines). Next, use the following commands, then physically reboot the system:

sudo apt update

sudo apt dist-upgrade

sudo apt install --fix-broken -o Dpkg::Options::="--force-overwrite"

If you encounter an error, please use the following 3 commands and re-attempt the upgrade (do not reboot the device after running below command and before re-attempting the upgrade)

sudo parted /dev/mmcblk0 -s rm 4

sudo parted /dev/mmcblk0 -s resizepart 3 67.9MB

sudo parted /dev/mmcblk0 mkpart -s A_reserved_on_user 67.9MB 101MB

Please select “Ignore” when prompted and “ext2” as the file system type.

Resources:

Note: Some of the links in Jetson Linux 35.2.1 page in Downloads section do not follow the normal convention. For example: https://developer.nvidia.com/downloads/r35releasev21-sources-ubuntufocal-l4taarch64srctbz2

Instead of

https://developer.nvidia.com/downloads/r35_release_v2.1/sources/ubuntu_focal-l4t_aarch64_src.tbz2

This is due to an issue in our posting tool and we plan to fix it within a week so that our links can follow the convention as before.

2 Likes
3

Update:

We found an issue when installing JetPack using SDK Manager where Jetson Multimedia installation was not succesfull. We have fixed the issue.
If you had already tried JetPack 5.1 installation using SDK Manager, please uninstall JetPack 5.1 using the SDK Manager menu in the left-bottom in the SDK Manager main GUI. Then close and re-open the SDK Manager, download and install JetPack 5.1 again.

Also note that the nvidia-jetpack metapackage is now avaialable in the APT repo. If you want to install JetPack after installing Jetson Linux 35.2.1 using APT, then just follow the familiar commands:

sudo apt update
sudo apt install nvidia-jetpack

5

Could you provide documentation on how to enable Temporal Adaptive Quantization in the video encoder?

6

Update:

Most of the containers for JetPack 5.1 are also up on NGC, except Tensorflow, Pytorch and ML container.

8

The ML containers for JetPack 5.1 have been published on NGC here:

nvcr.io/nvidia/l4t-ml:r35.2.1-py3
nvcr.io/nvidia/l4t-pytorch:r35.2.1-pth2.0-py3
nvcr.io/nvidia/l4t-tensorflow:r35.2.1-tf2.11-py3
9

Hello, I’m trying to flash the Jetson Orin NX, but it does not appear in the Target Hardware (SDKManager 1.9.1.10844 x86_64). Do you have any idea on how to solve this issue?

1 Like
10

Hi, we are looking for the fix for the [pthread_setschedparam] issue as seen in the patches here: Jetson/L4T/r35.1.x patches - eLinux.org

We are hoping to set our applications to SCHED_RR (round robin real time scheduling) with pthread_setschedparam. But as noted in many other discussions from recently and from a long time ago, e.g. these: pthread_setschedparam (SCHED_FIFO) fails and SCHED_RR on AGX Orin fails with Operation Not Permitted

… the only known workaround is to sysctl -w kernel.sched_rt_runtime_us=-1.

This workaround causes some problems. It starves other processes such as the one delivering video frames from our Realsense camera.

We tested on Jetpack 5.0.2 and 5.1 and the above patches list seems to indicate that a fix for this issue (?) is present at least in 5.0.2, but we still encounter pthread_setschedparam returning 1 (a failure) on 5.0.2 as well as 5.1. It seems to happen randomly.

is sysctl -w kernel.sched_rt_runtime_us=-1 the only workaround? Has anyone encountered any other alternatives?

In the mean time it seems like our only alternative is to re-engineer our apps to guarantee completing work enough to leave some CPU budget left for the rest of the processes on the system. Although this may be possible, it remains problematic that this pressure relief valve built into Linux cannot be used.

11

Hi, there’s a dedicated topic here: Orin NX not recognized when in recovery mode
but there does not seem to be a decent support of Orin NX with SDK Manager at the moment. No date is announced yet.

12

Hi,
Just wanted to drop a comment for those having the error: “Error: Can’t have the end before the start! (start sector=131112 length=-252)” with the second parted command. Fix is:

sudo parted /dev/mmcblk0 -s resizepart 3 67,9MB
sudo parted /dev/mmcblk0 mkpart -s A_reserved_on_user 67,9MB 101MB

Note the use of the comma instead of the dot.

Cheers

1 Like
13

Hi
I had older jetpack installed and moved the OS to nvme ssd.
when attempting to upgrade i get an error when updating l4t-core:
nvidia-l4t-core_35.2.1 sfdisk: partition 1: failed to get partition name

this causes unrecoverable broken packages
the same flow is working with Jetpack 5
Do you have any idea on how to solve this issue?

1 Like
18

My system accidentally got rebooted after the following command (it was running).

sudo parted /dev/mmcblk0 mkpart -s A_reserved_on_user 67.9MB 101MB

Now, my system doesn’t boot up. Do you have any idea on how to solve this problem?