参考Tensorflow Machine Leanrning Cookbook
The main role tf.ConfigProto () is arranged tf.Session operation mode, cpu or arithmetic operation such gpu
Specific code as follows:
import tensorflow as tf
session_config = tf.ConfigProto(
log_device_placement=True,
inter_op_parallelism_threads=0,
intra_op_parallelism_threads=0,
allow_soft_placement=True)
sess = tf.Session(config=session_config)
a = tf.constant([1.0, 2.0, 3.0, 4.0, 5.0, 6.0], shape=[2,3], name='b')
b = tf.constant([1.0, 2.0, 3.0, 4.0, 5.0, 6.0], shape=[3,2], name='b')
c = tf.matmul(a,b)
print(sess.run(c))
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Specific explanation
log_device_placement=True
When set to True, it will print out TensorFlow use the kind of operation
inter_op_parallelism_threads=0
A number of threads the thread provided inside the parallel arithmetic operations, such as matrix multiplication, if set to 0, the optimal number of threads handling
intra_op_parallelism_threads=0
A plurality of arithmetic operations in parallel the number of threads, such as c = a + b, d = e + f. Parallel operation
allow_soft_placement=True
Sometimes, different devices, it's cpu and gpu are different, if this option is set to True, the device does not meet the requirements when running, will automatically allocate GPU or CPU.
other options
When using the GPU time, Tensorflow automatically run slowly for maximum GPU memory
session_config.gpu_options.allow_growth = True
1
When a GPU, GPU memory maximum set ratio
session_config.gpu_options.per_process_gpu_memory_fraction = 0.4
1
Ability to use the GPU for computing
tf.test.is_built_with_cuda()
1
Additional processing method
import tensorflow as tf
sess = tf.Session()
with tf.device('/cpu:0'):
a = tf.constant([1.0, 3.0, 5.0], shape=[1, 3])
b = tf.constant([2.0, 4.0, 6.0], shape=[3, 1])
with tf.device('/gpu:0'):
c = tf.matmul(a, b)
c = tf.reshape(c, [-1])
with tf.device('/gpu:0'):
d = tf.matmul(b, a)
flat_d = tf.reshape(d, [-1])
= tf.multiply Combined (c, flat_d)
Print (sess.run (Combined))
---------------------
Author: debris flows in a clear stream
Source: CSDN
original: https: //blog.csdn.net/qq_31261509/article/details/79746114
copyright: This article is a blogger original article, reproduced, please attach Bowen link!