Skip to content

LBNL Science IT Docs TensorFlow 2.1

GitHub/lbnl-science-it/docs-source

TensorFlow 2.1

Jupyter Notebook on Lawrencium GPU node

https://sites.google.com/a/lbl.gov/high-performance-computing-services-group/getting-started/jupyter-notebook

$ srun  -N 1 -p es1 -A $ACCOUNT -t 1:0:0 --gres=gpu:2 -n 4 -q es_normal --pty bash
$ module load ml/tensorflow/2.1.0-py37
$ start_jupyter.py

https://sites.google.com/a/lbl.gov/high-performance-computing-services-group/getting-started/remote-desktop

import sys
import numpy as np
import tensorflow as tf
from datetime import date
from datetime import datetime

print(sys.version)

3.7.4 (default, Aug 13 2019, 20:35:49) [GCC 7.3.0]

print(tf.compat.v1.VERSION)

2.1.0

print(date.today())

2020-05-13

Test if TF can access a GPU

tf.config.list_physical_devices('GPU')

[PhysicalDevice(name='/physical_device:GPU:0', device_type='GPU'),
 PhysicalDevice(name='/physical_device:GPU:1', device_type='GPU'),
 PhysicalDevice(name='/physical_device:GPU:2', device_type='GPU'),
 PhysicalDevice(name='/physical_device:GPU:3', device_type='GPU')]

Print the name of the GPU device

tf.test.gpu_device_name()

'/device:GPU:0'