How to run Tensorflow on CPU

When running TensorFlow on CPU, first ensure that the correct version of TensorFlow is installed. TensorFlow supports both CPU and GPU execution environments, but by default, if no GPU is detected in the system, TensorFlow automatically runs on CPU.

Install TensorFlow

1. Install Python: TensorFlow requires a Python environment; it is recommended to use Python versions between 3.5 and 3.8.

2. Create a Virtual Environment (Optional): Using a virtual environment can avoid dependency conflicts and create an isolated environment for TensorFlow. You can use venv (built-in Python) or conda (Anaconda suite) to create a virtual environment.

   bash
   # Using venv
   python -m venv tf_env
   source tf_env/bin/activate  # Use tf_env\Scripts\activate on Windows
   
   # Using conda
   conda create -n tf_env python=3.8
   conda activate tf_env

3. Install TensorFlow: Install TensorFlow using pip. To ensure it runs on CPU, directly install the tensorflow package instead of tensorflow-gpu.

   bash
   pip install tensorflow

Verify Installation

After installation, verify that TensorFlow is correctly installed and runs on CPU by running a simple TensorFlow program.

python
import tensorflow as tf

# Create a TensorFlow constant
hello = tf.constant('Hello, TensorFlow!')

# Start TF session
tf_session = tf.compat.v1.Session()

# Run session to get results
print(tf_session.run(hello))

Configure TensorFlow to Use CPU

Although TensorFlow automatically runs on CPU, you may need to explicitly configure it to use only CPU, especially when the system has both CPU and GPU. This can be achieved by setting environment variables or configuring within the code.

python
import os
import tensorflow as tf

os.environ['CUDA_VISIBLE_DEVICES'] = '-1'  # This tells TensorFlow to ignore GPU

# Create some TensorFlow operations
a = tf.constant([[1.0, 2.0], [3.0, 4.0]])
b = tf.constant([[1.0, 1.0], [0.0, 1.0]])
c = tf.matmul(a, b)

print(c)

Example

For example, try using the CPU version of TensorFlow to implement a simple linear model.

python
import tensorflow as tf
import numpy as np

# Model and data parameters
X_data = np.array([[1.0, 2.0], [2.0, 3.0], [3.0, 4.0], [4.0, 5.0]])
y_data = np.array([[1.0], [2.0], [3.0], [4.0]])

# Build the model
X = tf.placeholder(tf.float32, shape=[None, 2])
y = tf.placeholder(tf.float32, shape=[None, 1])
W = tf.Variable(tf.random_normal([2, 1]), name='weight')
b = tf.Variable(tf.random_normal([1]), name='bias')
hypothesis = tf.matmul(X, W) + b

# Loss function and optimizer
cost = tf.reduce_mean(tf.square(hypothesis - y))
optimizer = tf.train.GradientDescentOptimizer(learning_rate=0.01).minimize(cost)

# Initialize variables
init = tf.global_variables_initializer()

# Start session
with tf.Session() as sess:
    sess.run(init)

    # Train the model
    for step in range(2000):
        cost_val, hy_val, _ = sess.run([cost, hypothesis, optimizer], feed_dict={X: X_data, y: y_data})
        if step % 500 == 0:
            print(step, "Cost:", cost_val, "\nPrediction:\n", hy_val)

The above example demonstrates how to create and train a simple linear regression model using TensorFlow on CPU. These steps ensure that TensorFlow effectively runs on CPU and processes data.