0
  1. follow the instruction from https://software.intel.com/en-us/articles/build-and-install-tensorflow-on-intel-architecture
  2. bazel build --config=mkl --copt="-DEIGEN_USE_VML" -s -c opt //tensorflow/tools/pip_package:build_pip_package
  3. bazel-bin/tensorflow/tools/pip_package/build_pip_package /tmp/tensorflow_pkg
  4. pip install /tmp/tensorflow_pkg/tensorflow-1.2.1-cp27-cp27mu-linux_x86_64.whl

When I import tensorflow, got the warning

2017-08-16 15:30:32.132057: E tensorflow/core/framework/op_kernel.cc:1108] OpKernel ('op: "Fact" device_type: "CPU" label: "sergey"') for unknown op: Fact

2017-08-16 15:30:32.132104: E tensorflow/core/framework/op_kernel.cc:1108] OpKernel ('op: "Fact" device_type: "GPU" host_memory_arg: "fact"') for unknown op: Fact

2017-08-16 15:30:32.132689: E tensorflow/core/framework/op_kernel.cc:1108] OpKernel ('op: "Fact" device_type: "CPU" label: "Sergey"') for unknown op: Fact WARNING:tensorflow:From /home/admin/anaconda2/lib/python2.7/site-packages/tensorflow/python/util/tf_should_use.py:170: initialize_all_variables (from tensorflow.python.ops.variables) is deprecated and will be removed after 2017-03-02. Instructions for updating: Use tf.global_variables_initializer instead.

2017-08-16 15:30:32.262141: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use SSE4.1 instructions, but these are available on your machine and could speed up CPU computations.

2017-08-16 15:30:32.262189: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use SSE4.2 instructions, but these are available on your machine and could speed up CPU computations.

2017-08-16 15:30:32.262203: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use AVX instructions, but these are available on your machine and could speed up CPU computations.

2017-08-16 15:30:32.262222: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use AVX2 instructions, but these are available on your machine and could speed up CPU computations.

2017-08-16 15:30:32.262239: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use FMA instructions, but these are available on your machine and could speed up CPU computations.

Does anyone get the similar warning? and how to fix it. BTW, i use anaconda (conda 4.0.5) to install the tensorflow package.

GPhilo
  • 18,519
  • 9
  • 63
  • 89
user2223740
  • 31
  • 3
  • 1
    Take a look at this question and the first answer: [https://stackoverflow.com/questions/41293077/how-to-compile-tensorflow-with-sse4-2-and-avx-instructions](https://stackoverflow.com/questions/41293077/how-to-compile-tensorflow-with-sse4-2-and-avx-instructions), you might need to add the flags `--copt=-mavx --copt=-mavx2` to your bazel build command – GPhilo Aug 16 '17 at 14:59
  • Actually, thanks for your comment, problem resolved. – user2223740 Aug 17 '17 at 10:44
  • Glad to hear it helped! You might want to add and accept your own answer detailing which options you used in the build to make it work, it sounds like a problem other people might run into – GPhilo Aug 17 '17 at 10:45
  • 1
    Possible duplicate of [How to compile Tensorflow with SSE4.2 and AVX instructions?](https://stackoverflow.com/questions/41293077/how-to-compile-tensorflow-with-sse4-2-and-avx-instructions) – Peter Cordes Sep 02 '17 at 22:54
  • `-mavx2` doesn't enable FMA. You should use `--copt=-march=native`. – Peter Cordes Sep 02 '17 at 22:55

0 Answers0