Making sure CUDA is installed in Jetson Nano

Type in nvcc –version from your Jetson Nano terminal is the sure way.

If the nvcc is missing, it could mean that the jetson-toolkit were not installed, or the binaries was not added to your session.

To reassociate your user to installed nvcc

The commands to reassociate.

Check for CUDA default location at /usr/local

ls -larth /usr/local/

Next based on the returned directory run the following command

echo 'export PATH=/usr/local/cuda/bin:$PATH' >> ~/.bashrc
echo 'export LD_LIBRARY_PATH=/usr/local/cuda/lib64:$LD_LIBRARY_PATH' >> ~/.bashrc

The command will add the path to your user shell session.

Then, rerun the nvcc –version

Next run the utiliy tegrastats

Result of the tegrastats

Further test will be running some of the example, after further queries and forums.

Here are the simple examples from nVidia CUDA Samples :: CUDA Toolkit Documentation

Command in sequence are:

cd /usr/local/cuda/samples/1_Utilities/deviceQuery
sudo make
./deviceQuery
Results of compiling C sample for the deviceQuery
Full result of the deviceQuery

If you are able to get the same result from this post, it is an indication that your CUDA is installed properly on your Jetson Nano.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.