Skip to content

Instantly share code, notes, and snippets.

@xsqian
Last active August 13, 2021 04:26
Show Gist options
  • Save xsqian/8e182d543d684d92d99d6bb86951cad4 to your computer and use it in GitHub Desktop.
Save xsqian/8e182d543d684d92d99d6bb86951cad4 to your computer and use it in GitHub Desktop.
install tensorflow with GPU support
#conda create an env
conda create -n py376-tf241 ipykernel python=3.7.6
conda activate py376-tf241
# find out current gpu avaliablity
import tensorflow as tf
print(tf.__version__)
print(tf.config.list_pysical_devices())
# cudnn and cudatoolkit
conda list | grep cuddnn
conda list | grep cudatoolkit
# cuda compiler version
nvcc -V
#specific versionof cudnn and cudatoolkit
conda search cudnn
# install the specific version
conda install cudnn=7.6.5=cuda10.1_0
#install gpu compatible tensorflow
conda install tensorflow-gpu
#check the tf and gpu again
import tensorflow as tf
print(tf.__version__)
print(tf.config.list_physical_devices())
pip install numba
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment