Skip to content

TensorFlow 1.12

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/1.12.0-py36
$ start_jupyter.py

Login from visualization node via VNC Viewer:

  • 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)

3.6.8 |Anaconda custom (64-bit)| (default, Dec 30 2018, 01:22:34) [GCC 7.3.0]

print(tf.VERSION)

1.12.0

print(date.today())

2020-04-30

Test if TF can access a GPU

tf.test.is_gpu_available(
    cuda_only=False,
    min_cuda_compute_capability=None
)

True

tf.test.gpu_device_name()

'/device:GPU:0'

Run the code on a GPU or CPU

device_name = tf.test.gpu_device_name()
#device_name = '/device:CPU:0'
shape = (3000, 3000)
with tf.device(device_name):
    random_matrix = tf.random_uniform(shape=shape, minval=0, maxval=1)
    dot_operation = tf.matmul(random_matrix, tf.transpose(random_matrix))
    sum_operation = tf.reduce_sum(dot_operation)
startTime = datetime.now()
with tf.Session(config=tf.ConfigProto(log_device_placement=True)) as session:
        result = session.run(sum_operation)
        print(result)

print("Shape:", shape, "Device:", device_name)
print("Time taken:", datetime.now() - startTime)

6750466000.0 Shape: (3000, 3000) Device: /device:GPU:0 Time taken: 0:00:00.044407

#device_name = tf.test.gpu_device_name()
device_name = '/device:CPU:0'
shape = (3000, 3000)
with tf.device(device_name):
    random_matrix = tf.random_uniform(shape=shape, minval=0, maxval=1)
    dot_operation = tf.matmul(random_matrix, tf.transpose(random_matrix))
    sum_operation = tf.reduce_sum(dot_operation)
startTime = datetime.now()
with tf.Session(config=tf.ConfigProto(log_device_placement=True)) as session:
        result = session.run(sum_operation)
        print(result)

print("Shape:", shape, "Device:", device_name)
print("Time taken:", datetime.now() - startTime)

6750241300.0 Shape: (3000, 3000) Device: /device:CPU:0 Time taken: 0:00:00.435257