I recently bought a jetson orin nano and am trying to run a tensorflow model on it. after going throught the installation process from the tensorflow website (tensorflow[and-cuda]), the command “tf.config.list_physical_devices(‘GPU’)” is an empty list, any fixes?
Hi,
You can build it from the source with the repo below:
Thanks.
Hello
Thank you for the fast reply.
Can it be done outside of a container or is this the only way?
Hi,
You can also build the package natively without using a container.
The steps can be found in the script below:
#!/usr/bin/env bash
# TensorFlow builder for Jetson (architecture: ARM64, CUDA support)
set -ex
echo "Building Tensorflow ${TENSORFLOW_VERSION}"
# Install LLVM/Clang 20 # Compatible with blackwell
wget https://apt.llvm.org/llvm.sh
chmod u+x llvm.sh
./llvm.sh 20 all
echo "Building TensorFlow for Jetson"
# Clone the TensorFlow repository
git clone --branch "v${TENSORFLOW_VERSION}" --depth=1 https://github.com/tensorflow/tensorflow.git /opt/tensorflow || \
git clone --depth=1 https://github.com/tensorflow/tensorflow.git /opt/tensorflow
cd /opt/tensorflow
# Set up environment variables for the configure script
export HERMETIC_PYTHON_VERSION="${PYTHON_VERSION_MAJOR}.${PYTHON_VERSION_MINOR}"
This file has been truncated. show original
Thanks.
system
Closed
August 13, 2025, 1:51am
7
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.