SDK Manager not working


I would like to inform that the SDK Manager for installing Orin does not work properly.
After 5 attempts to flash and after install the Cuda etc and deepstream it does not work with USB-connection.
The flashing works but not the USB-connection…
I installed it with IP after a lot of hassle and managed to (seemingly) get Deepstream installed.
Now I am stuck on the pyds package and why the "¤#%& dont you provide it with the installation of deepstream on Orin. The python community is huge and if you sell this SDK as End to end and so fantastic, then you better make tutorials that actually can be followed. There are now, as I learn as I go along, that there are a lot of prerequisites to install that are linked like a spiderweb. A suggestion is to look at the communities around AI and your tools to understand the level of instructions to give. Paul McWorther is one whose videos are well designed.

From your another post at DeepSteam SDK forum - Deepstreem installation fails - Intelligent Video Analytics / DeepStream SDK - NVIDIA Developer Forums
That seems the installation work.
What’s your host PC OS version to install SDKM?


The post you refer to is mine too. And after that hassle and also the problems I have had with Orin the first time I wanted to let you know that the SDK-manager is really unreliable.

I downloaded the latest SDKM and installed on Ubuntu 18.04.
The first thing that happened when trying to update my SDKM was an error so I had to install the package manually and not use the built in update mangager for the SDKM.

I would like to see a video where you show the steps to install. Especially when swapping between flashing and SDK-installation. The flashing part worked most of the times but the installation of SDK did not.

Also. When using deepstream for Jetson I would prefer that the Python bindings are included since its quite a crucial part for that platform. It is also difficult to get the thing to work.

Please also add an open source data stream and video file with h.264 to use as sample so that you can really test that the installation has worked and all the bits fits.

For Jetson Orin, the Ubuntu 20.04 is suggested with the latest SDK manager.
You can check with Getting Started with the NVIDIA Jetson AGX Orin Developer Kit - YouTube
and Getting Started | NVIDIA Developer

Good info thanks.

Reference from
GitHub - dusty-nv/jetson-inference: Hello AI World guide to deploying deep-learning inference networks and deep vision primitives with TensorRT and NVIDIA Jetson. to: Page Not Found | NVIDIA

Link does not exist.
The resources does not include deepstream which should be simple and easy to use and deploy AI.
I know you are broad and describe this in depth but I still feel you miss out on the audience who
would use it if it was easier to find on the website. It is ALOT of info and references and prerequisites.
The benefit, REALLY, with Jetson and your solution is the simplicity and accessability of AI implementation to the masses. But in this case, and the criticism I give, is two things. 1. Tricky SDK that is not working properly.
2. Easy access to “Hello” world for deepstream.

So conclusion. When getting Orin, Flash with the latest OS and SDK, I would expect that the system would work and yet I get immediately stuck on linux/python packs that does not install or that I have to compile from source. I would expect more. I will try SDKM for Ubuntu 20 and see if that works better.

Hi @magnus.gabell, thanks for your valuable feedback and sorry for the frustrations that you’ve been having - I’ll discuss this with the DeepStream team about the Python bindings and samples. We previously had a good introductory DLI course for DeepStream/IVA which included a container for JetPack 4.x that had everything in it ready-to-run, but we’ve been needing to update that for JetPack 5.x.

In the meantime, please continue following-up with @kayccc about any issues you are facing with SDK Manager and we’ll be happy to help.

Thank you! I will do. And for the record, your support is excellent and excellent response time. Thanks for the good work!

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.