Tensorflow set CUDA VISIBLE DEVICES within jupyter

Question

I have two GPUs and would like to run two different networks via ipynb simultaneously  however the first notebook always allocates both GPUs    Using CUDA VISIBLE DEVICES  I can hide devices for python files  however I am unsure of how to do so within a notebook   Is there anyway to hide different GPUs in to notebooks running on the same server

User · Answer

You can also enable multiple GPU cores  like so  import os os environ  quot CUDA DEVICE ORDER quot    quot PCI BUS ID quot  os environ  quot CUDA VISIBLE DEVICES quot    quot 0 2 3 4 quot

User · Answer

You can do it faster without any imports just by using magics    env CUDA DEVICE ORDER PCI BUS ID  env CUDA VISIBLE DEVICES 0   Notice that all env variable are strings  so no need to use    You can verify that env-variable is set up by running   env  lt name of var gt   Or check all of them with  env

User · Answer

You can set environment variables in the notebook using os environ  Do the following before initializing TensorFlow to limit TensorFlow to first GPU   import os os environ  CUDA DEVICE ORDER    PCI BUS ID      see issue  152 os environ  CUDA VISIBLE DEVICES    0    You can double check that you have the correct devices visible to TF  from tensorflow python client import device lib print device lib list local devices     I tend to use it from utility module like notebook util  import notebook util notebook util pick gpu lowest memory   import tensorflow as tf

[python] Tensorflow set CUDA_VISIBLE_DEVICES within jupyter

Examples related to python

Examples related to environment-variables

Examples related to tensorflow