TensorFlow.js: How to avoid `Your CPU supports instructions ... AVX AVX2`?

user_9494636 0
Viewed 8340

Environment:

  • Windows 10 x64,
  • Node.js v10,
  • @tensorflow/tfjs-node v0.1.15

I'm trying to use tensorflow.js on Node.js.

I installed tfjs-node, and it auto-built successfully (node-gyp), but I receive the following error when running:

tensorflow/core/platform/cpu_feature_guard.cc:141] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX AVX2

The similar question in Python version can be found here:

Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX AVX2

Currently, I don't care about performance, so I just want to disables the warning, don't enable AVX/FMA. In JavaScript, what should I do?

tensorflow javascript
2 Answers

Set environment variables before running.

Windows:

$ set TF_CPP_MIN_LOG_LEVEL=2

Linux/MacOS:

$ export TF_CPP_MIN_LOG_LEVEL=2

user_94946360

First list out the packages (Tensorflow and related) that you have using conda list command.

>> conda list
This command output with below tensorflow version
tblib                     1.3.2            py36h30f5020_0
tensorboard               1.13.1                    <pip>
tensorflow                1.13.1                    <pip>
tensorflow-estimator      1.13.0                    <pip>

Note: I have omitted some of the unwanted listing of packages

Installing collected packages: tensorflow Successfully installed tensorflow-1.13.1
You are using pip version 18.0, however version 19.0.3 is available.
You should consider upgrading via the 'python -m pip install --upgrade pip' command.

(base) C:\Users\shashi>python
Python 3.6.6 |Anaconda custom (64-bit)| (default, Jun 28 2018, 11:27:44) [MSC v.1900 
64 bit (AMD64)] on win32
Type "help", "copyright", "credits" or "license" for more information.
>>> import tensorflow as tf
>>> hello = tf.constant('Hello Tensorflow')
>>> sess = tf.Session()
I tensorflow/core/platform/cpu_feature_guard.cc:141] Your CPU supports instructions 
that this TensorFlow binary was not compiled to use: AVX2
>>> import os
>>> import os
>>> os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2'
>>> hello = tf.constant('Hello Tensorflow')
>>> sess = tf.Session()
>>> print(sess.run(hello))
b'Hello Tensorflow'

After the value of os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' has been set, the code works fine.

If Jupyter also throws error, then close the opened Jupyter notebooks and again re-open them. It will all work fine.

user_105274030
Related