[BUG]: "bash: nvcc: command not found" On Jetson Nano (4GB)

[mericgeren]

Is this a duplicate?

• I confirmed there appear to be no duplicate issues for this bug and that I agree to the Code of Conduct

Type of Bug

Something else

Component

Not sure

Describe the bug

Hello,

I have installed Jetpack 4.6.4 from quick start guide with micro sd on a Jetson Nano recently. After the installation i had 3 cuda folders inside usr/local named "cuda", "cuda-10" and "cuda-10.2" and nvcc was working. Then, i tried to install and build OpenCV with CUDA support following these steps specified here under " Installation script." section. The build have failed. Then, when i checked "usr/local" and saw there is only "cuda-10.2" . After that, i tried nvcc --version and saw:

bash: nvcc: command not found

After that, created a link with sudo ln -sfT /usr/local/cuda-10.2/ /usr/local/cuda command. Then, checked if it works with ls -l /usr/local/cuda command and got this on my terminal:

lrwxrwxrwx 1 root root 21 Sep 13 11:46 /usr/local/cuda -> /usr/local/cuda-10.2/

Then, checked my .bashrc file and there were these lines at the end of .bashrc:

 export PATH=/usr/local/cuda/bin${PATH:+:${PATH}}
 export LD_LIBRARY_PATH=/usr/local/cuda/lib64${LD_LIBRARY_PATH:+:${LD_LIBRARY_PATH}}

Then tried again the command: nvcc --version and got bash: nvcc: command not found on my terminal again. Later on, i checked with jtop and saw these lines at the "info" section:

Libraries

cuDNN: 8.2.1.32

TensorRT: 8.2.1.8

CUDA: MISSING

Vulkan: 1.2.70

VPI: 1.2.3

OpenCV: 4.1.1 with CUDA: NO

How to Reproduce

1. Flash JetPack to Jetson Nano with micro sd method.
2. Be sure that these two lines are present at the end of .bashrc:

export PATH=/usr/local/cuda/bin${PATH:+:${PATH}}
export LD_LIBRARY_PATH=/usr/local/cuda/lib64${LD_LIBRARY_PATH:+:${LD_LIBRARY_PATH}}

3. Follow steps specified here under "Installation script."
4. See that line: "-- Configuring incomplete, errors occurred!"
5. Check "/usr/local" and see that there is only "cuda-10.2" as cuda folder.
6. Enter the command nvcc --version and see "bash: nvcc: command not found
   "
7. Check status of CUDA with "jtop" and see that CUDA is missing.

Expected behavior

Seeing version of nvcc when entering the command nvcc --version and seeing CUDA version when checking with "jtop".

Reproduction link

No response

Operating System

L4T 32.7.4 (Based on Linux 18.04)

nvidia-smi output

Not available. Since, i am using Jetson Nano.

NVCC version

The issue is not being able to check it.

[github-actions]

Hi @mericgeren!

Thanks for submitting this issue - the CCCL team has been notified and we'll get back to you as soon as we can!
In the mean time, feel free to add any relevant information to this issue.

[miscco]

Hi @mericgeren,

I fail to see how this is a cccl issue? We are not responsible for OpenCV installation documentation.

Furthermore, I am afraid that the installed 10.2 version is not supported, as we require at least 11.0.

Anyway you can always download the cuda toolkit from here https://developer.nvidia.com/cuda-downloads

[mericgeren]

Hi @miscco

The issue is even though i added directories for cuda in ".bashrc" and linked "cuda" and "cuda-10.2" (which is currently installed one) on Jetson Nano (for which CUDA is included in Jetpack) failing to find the installed cuda in the system. If it wouldn't be an issue about CUDA i wouldn't think about reporting here. But, if i am not mistaken it's an issue with CUDA with links to the context which includes the OpenCV installation i have mentioned.

[miscco]

Just to be clear, we are the cccl project.

Our libraries cub, thrust and libcu++ are used to develop c++ programs using cuda. We are indeed packaged with the cuda toolchain, but we have no business regarding its installation.

For jetson Nano, I can only refer you to the official site https://developer.nvidia.com/embedded/learn/get-started-jetson-nano-devkit

[jarmak-nv]

Hey @mericgeren

You may find better success if you post a question on the Jetson Dev Forums here: https://forums.developer.nvidia.com/c/agx-autonomous-machines/jetson-embedded-systems/jetson-nano/76

Quick digging into your links also pointed me here https://github.com/Qengineering/Jetson-Nano-image where they provide an image that might get you started quickly. They are probably better equipped to help you as well.

I hope you can resolve the issue - good luck!