How to install cudnn & tensorflow gpu on jetson nano

am trying to run

On jetson nano

downloaded tf_gpu-1.13.1+nv19.5-py3
from
https://developer.nvidia.com/embedded/downloads#?search=Tensorflow

tried to install on jetson nano. failed.

Looks like cuDNN is not installed. please guide me how to install

Hi,

Please install CUDA, cuDNN and TensorRT before installing the TensorFlow package.
You can install all the requirements from the JetPack installer:
[url]https://developer.nvidia.com/embedded/jetpack[/url]

Thanks.

Hi,

Thanks,

CUDA, cuDNN and TensorRT are installed through JetPack.

I want to run the following on jetson nano

https://github.com/stereolabs/zed-tensorflow/tree/5b91c59482794f53d1d88e5877b0e1a3e83c098f

when I run the following

pip3 install --upgrade tensorflow-gpu

getting an error
could not find a version that satisfies the requirement …
no matching distribution found for tensorflow-gpu

is it possible to run this sample on Jetson Nano? How is it done?

Hi,

The official TensorFlow doesn’t support ARM platform.
But we release several prebuilt package for the Jetson users.

Please find the installation commands in this topic:
[url]https://devtalk.nvidia.com/default/topic/1048776/jetson-nano/official-tensorflow-for-jetson-nano-/[/url]

Thanks.

hello, I used jetpack on my nano, but I didn’t find cudnn. So I want to install cudnn by myself, but I could not find a cudnn version in aarch64. Thank you

1 Like

Hi,

All the ARM-based packages are included in the JetPack installer.
Please install cuDNN package directly from the JetPack.

After the installation, you should be able to find the cuDNN sample at /usr/src/cudnn_samples_v7/.
Thanks.