JetPack L4T 3.2 Release fails to install CUDA Toolkit on host machine.

Robotics & Edge Computing Jetson & Embedded Systems Jetson TX2
21

Hello
I am having the exact same problem. Has there been found any solutions to this?

These are the last lines of the installation log

E: Failed to fetch http://no.archive.ubuntu.com/ubuntu/dists/xenial-backports/main/binary-arm64/Packages  404  Not Found [IP: 193.35.52.51 80]
E: Some index files failed to download. They have been ignored, or old ones used instead.
Reading package lists...
Building dependency tree...
Reading state information...
E: Unable to locate package cuda-toolkit-8-0-82
E: Unable to locate package cuda-cross-aarch64-8-0-82

I too had CUDA 9.1 installed, but removed it with apt-get remove.

It seems like the Jetpack installer is looking for the 8.0.82 version from the log, but in my /var folder the 8.0.64 version is added during installation in the folder “cuda-repo-8.0.local”. All the files inside this folder has names with “…8.0.64…” in them. Can this have something to do with the issue?

22

Hi,

We want to confirm your issue first:
Do you meet package error when installing CUDA toolkit on an x86 environment with JetPack?

Different pre-installed CUDA version may make JetPack confuse about the current status.
Could you remove all the CUDA toolkit first.
And delete the files and logs generated from JetPack and run it again?

Thanks.

23

I was unable to resolve the issue after uninstalling JetPack and removing all CUDA packages. Had to reinstall a fresh ubuntu version for the issue to be resolved

24

There is a bug in JetPack3.2 ever since JetPack3.2.1 came out. The downloaded repository.json has reverted to an older version, so you should instead use JetPack3.2.1. The repository.json file gets downloaded each time JetPack is run, and content may vary depending on which download server you go to, but this shows you’ve hit this bug (NVIDIA servers need to be updated…deprecation of JetPack3.2 was incomplete or incorrect):

E: Unable to locate package cuda-toolkit-8-0-82

JetPack3.2 uses CUDA 9, not CUDA 8, so it shouldn’t even be looking at this. Each new run has a chance to fix it if the server provides the correct repository.json.

25

Thanks for the explanation. I will keep that in mind if I run into further issues