Looking for a pre-built wheel for PyTorch for Jetson Nano

Robotics & Edge Computing Jetson & Embedded Systems Jetson Nano
, ,
1

I’m trying to install PyTorch with CUDA support on a Jetson Nano. It’s running the Ubuntu 18.04 that the jetson-nano-jp461-sd-card-image image from Nvidia put on it with Python 3.6.9, CUDA 10.2, and JetPack 4.6.6. I tried:

pip3 install torch --index-url https://download.pytorch.org/whl/cpu

but this gives me the CPU-only version, and I need CUDA support for the Nano’s GPU. I also tried

git clone --recursive --branch v1.8.0 GitHub - pytorch/pytorch: Tensors and Dynamic neural networks in Python with strong GPU acceleration

and installing, but it is really, really, slow.

Is there a pre-built wheel available for this specific configuration? Many thanks in advance.

3

Hi,

Yes, please find the package in the below link:

Thanks.

4

Thanks so much! I can confirm that for my setup, this worked:
sudo apt-get install python3-pip libopenblas-base libopenmpi-dev libomp-dev
pip3 install ‘Cython<3’
pip3 install numpy
wget https://nvidia.box.com/shared/static/fjtbno0vpo676a25cgvuqc1wty0fkkg6.whl -O torch-1.10.0-cp36-cp36m-linux_aarch64.whl
sudo pip3 install torch-1.10.0-cp36-cp36m-linux_aarch64.whl