i installed jetpack 3.3.2 in my jetson tx2. the target components are not installed in my tx2, but the sdk manager shows that all the components have been installed . kindly please help me to rectify the error im facing…
Please help check below two path to make sure sdk components install success or not:
$ ll /usr/local/cuda
$ ll /home/nvidia/
The latest JetPack version is JetPack-4.3.
Suggest you can using sdkmanager to install again on your TX2.
is there any other jetpack which uses ubuntu 16.04…??
All the JetPack can run on Ubuntu 16.04/18.04 host machine.
Note: You can only use the JetPack which flashed the unit to install extra components. You mentioned JetPack 3.3.2, so that should be used (and as @carolyuu mentioned, this works from a host PC, and Ubuntu 16.04 will work for that PC…you don’t need to flash again, just deselect flash and select to add components to the Jetson while the Jetson is fully running and available by ethernet).
i’ve tried that also but it ain’t working …
Can you give the error message? Is it a particular package?
no samples are available and no packages are found example cuda packages,computer vision packages and etc…
Normally JetPack/SDK Manager will download a manifest of relevant files for your NVIDIA login and the release/hardware being considered. Then actual packages are downloaded via that manifest list. If NVIDIA login fails, then the manifest won’t exist, and if the network to NVIDIA fails, there also won’t be a manifest. If the URLs for the manifest have network issues, e.g., firewalls or proxies, then this too would cause failure to download.
Just to check your login, can you go here and log in normally?
If that works, then within JetPack/SDK Manager you should also be able to log in (just make sure you are logging in with the correct “developer.nvidia.com” tab (the “partners.nvidia.com” would be a different login). If that works, then you know you can obtain the manifest. Any failures after this are probably issues with a route to download servers. Knowing where it is failing would help debug.
Incidentally, CUDA is the base requirement for almost everything. Uncheck everything except CUDA to start testing. Samples get more complicated because they are cross compiled on the host PC and add requirements to the host PC itself.