AVX2

[jihaonew]

Did you use AVX2 flag when build gpu version with python3.6 ?

[danqing]

Yes. Are you seeing otherwise?

[jihaonew]

Yes. r1.4 is available now. BTW, I encountered an error when install tf 1.3, python 3.6. The error is about MPI. I installed mpich and got the same error. I didn't log that error down. I may install 1.3 in the next few days. After that, I will feedback here. Besides, can you provide caffe packge wheels?

[danqing]

If you don't need MPI, use the non-MPI version. Did you not see AVX2 or did you have issue with MPI?

1.4 will come once it's stable.

[danqing]

Re. caffe, are you talking about caffe 1 or caffe 2?

[danqing]

Closing due to inactivity.

TF 1.4 has been uploaded, and AVX2 is supported. Please feel free to reopen if you have further questions.

[jihaonew]

thanks. 发自网易邮箱大师 On 2017-11-11 02:40 , Danqing Liu Wrote: Closing due to inactivity. TF 1.4 has been uploaded, and AVX2 is supported. Please feel free to reopen if you have further questions. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or mute the thread.

[jax79sg]

Hi, i pip uninstalled tensorflow, sudo -H pip3 uninstall tensorflow and reinstall tensorflow using the whl file provided by the site. sudo -H pip3 install --upgrade --force-reinstall --no-index --find-links files://home/user/pip tensorflow==1.3.0

However i am still getting the SSE4.1, 4.2, AVX, AVX2...etc warnings after executing the following. Did i miss a step? import tensorflow as tf sess = tf.Session() Thanks.

[danqing]

@jax79sg please open a separate issue so we can help you. Essentially, you want to do something like

pip3 --no-cache-dir install https://github.com/mind/wheels/releases/download/tf1.3.1-cpu/tensorflow-1.3.1-cp36-cp36m-linux_x86_64.whl