search

chldydgh4687 / 00Self-Linux_trial_and_error

Linux & DeepLearning
0 stars 0 forks source link

[TIP] [Linux] Tensorflow_GPU 할당 #13

Open chldydgh4687 opened 4 years ago

chldydgh4687 commented 4 years ago

1.터미널에서 python code를 실행하는 경우: 

$ CUDA_VISIBLE_DEVICES=0 python script.py  
$ CUDA_VISIBLE_DEVICES=1 python script.py  
$ CUDA_VISIBLE_DEVICES=2,3 python script.py
  1. 코드에 직접 python code를 넣는 방법: 
    
import os

os.environ["CUDA_DEVICE_ORDER"]="PCI_BUS_ID"
os.environ["CUDA_VISIBLE_DEVICES"]="0"