- Jetson Nano B01
- JetPack 4.6.1 image burned on SD
- Manually installed python 3.7.5 (Need 3.7)
$ sudo apt-get install python3.7 python3.7-dev
# installed from a third party wheel
https://github.com/KumaTea/tensorflow-aarch64/releases/tag/v2.7
>>> import tensorflow as tf
>>> tf.config.list_physical_devices('GPU')
[]
My question:
Unable to find wheel for Python 3.7 from Nvidia(https://developer.download.nvidia.com/compute/redist/v461). How can I find a tensorflow wheel which will use gpu?
Thanks
Does the 3rd party wheel support GPU?
IF not, try to follow NVIDIA’s official instruction to install that.
Hi,
Our official package has CUDA support.
But we only support the default Python version.
For Nano, it’s Python 3.6:
https://docs.nvidia.com/deeplearning/frameworks/install-tf-jetson-platform/index.html
Thanks.
Hi,
I am aware that the official tensorflow package supports GPU.
However I need an approach to use tensorflow with python3.7 to support GPU.
I was wondering if anyone has done or needed it.
Thanks
Hi,
We only provide the Python3.6 TensorFlow package for Nano.
If you want to use Python 3.7, please build it from the source.
Thanks.
If you want to use Python 3.7, please build it from the source.
I would like to build it from the source to make it support GPU.
Could you point me to the right direction of the source?
Thanks
@AastaLLL
Could you help me find the correct source? Thanks
Hi,
The tensorFlow source can be found in the below link:
Thanks.
system
Closed
11
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.