TensorFlow 1.13.1 CPU (no SSE4.x, no AVX), Python 3.6, Ubuntu 18.04, amd64

[mdsimmo]

TF	OS	Py	SSE4.1	SSE4.2	AVX	AVX2	FMA	MKL
1.13.1	Ubuntu 18.04	3.6	:x:	:x:	:x:	:x:	:x:	:x:

compiled on: amd phenom II x4, 64 bit, Xubuntu 18.04, native compile flags

For those of you with older CPUs

Download: https://github.com/mdsimmo/tensorflow-community-wheels/releases/download/1.13.1_cpu_py3_6_amd64/tf_nightly-1.13.1-cp36-cp36m-linux_x86_64.whl

Install: sudo -H pip3 install https://github.com/mdsimmo/tensorflow-community-wheels/releases/download/1.13.1_cpu_py3_6_amd64/tf_nightly-1.13.1-cp36-cp36m-linux_x86_64.whl

Python 2.7 version: https://github.com/yaroslavvb/tensorflow-community-wheels/issues/97 Python 3.5 version: https://github.com/yaroslavvb/tensorflow-community-wheels/issues/97#issuecomment-475467617 GPU version: https://github.com/yaroslavvb/tensorflow-community-wheels/issues/107

If you have SSE4 instructions, then consider using apaid's build here: https://github.com/yaroslavvb/tensorflow-community-wheels/issues/97#issuecomment-459285105 (tf 1.12)

[evanatyourservice]

Perfect! Thank you!

[edaustin]

Could you build a GPU version of the above? It's sadly needed )))))

[mdsimmo]

@edaustin I can't as I don't have a compatible GPU to test with. You might have luck asking @davidenunes at https://github.com/davidenunes/tensorflow-wheels

[edaustin]

It's OK, have built it from scratch, please see my entry just added )) I couldn't damn well find one anywhere.. hopefully, others will also find it handy!

[mdsimmo]

Cool. For others looking, here's the link: https://github.com/yaroslavvb/tensorflow-community-wheels/issues/107

[iridiumblue]

You, sir, are an excellent human.

[andrei-q]

ImportError: /usr/lib/x86_64-linux-gnu/libstdc++.so.6: version `CXXABI_1.3.11' not found (required by /mnt/lin2/venv/p36/lib/python3.6/site-packages/tensorflow/python/_pywrap_tensorflow_internal.so) Failed to load the native TensorFlow runtime.

[mdsimmo]

@andrei-q Looks like your gcc is outdated. From here: https://gcc.gnu.org/onlinedocs/libstdc++/manual/abi.html (ctr f: "CXXABI_1.3.11"), it would seem you need at least gcc 7.1.0.

If you happen to be on a RaspPi or similar (I know they have old gcc installs), any reason why you wouldn't use Tensorflow Lite?

[TonyCongqianWang]

I installed this version, but when I use keras,

model = tf.keras.Sequential() model.add(Dense(256 * staticMult, activation="relu", input_shape=(X_train.shape[-1],)))

the first line works, but it gets stuck at the second line.

E: I just found out, that without input_shape it works fine. But with, it just gets stuck and cant be stopped by ctrl C. But then it gets stuck later at the compile step

[sona1111]

Ran into a similar issue as @TonyCongqianWang

[mdsimmo]

@TonyCongqianWang Agreed, I get same issue. Full code that it gets stuck on for me:

from keras.models import Sequential
from keras.layers import Dense

model = Sequential()
model.add(Dense(units=64, activation='relu', input_dim=100))

[mzalazar]

Tensorflow 1.13 built from source (after 8+- hours compiling) CPU ONLY version, compiled using: "bazel build --config=opt //tensorflow/tools/pip_package:build_pip_package" as stated in https://www.tensorflow.org/install/source Here is the result (package): https://mega.nz/#!kEVFlKgR!4LMfFJttn3KQmSyCK8Kv4RmzqEwALbYkuI5qUevG5lE

Install using: unzip tensorflow-1.13.2-cp37-cp37m-linux_x86_64.zip pip install tensorflow-1.13.2-cp37-cp37m-linux_x86_64.whl

My specs: Samsung RV511 CPU: Intel(R) Pentium(R) CPU P6200 @ 2.13GHz Linux Mint 19 Tara Python 3.7.5

Now... tensorflow is working in my old machine =)