Expert73
Expert73
Have you tried using mkl optimization? Does it give a performance boost? The same question about dnnl.
as far as I understand these builds are only for Linux, isn't it?
Thanks! Is automatic mixed precision supported by tf.keras in Tensorflow 2.0 (GPU & AVX2)?
Thanks for the answer!
Without CuDNNLSTM, performance dipped 3-5 times. CPU load increased 2 times (70% vs 35%). How to replace CuDNNLSTM correctly? LSTM is not a good replacement option. Please help.
> What is your problem? Can't you use `tf.compat.v1.keras.layers.CuDNNLSTM`?? All works. Thank you! but too slow. TF 1.14 (CPU load - 20%, GPU load - 20%, time 8 min) TF...
I have CPU with only SSE2 instruction. My GPU - NVIDIA 650Ti. i wrote my code 2 years ago. On older versions of the tensorflow that could be installed through...
if I want to compile thensorflow myself on CPU with sse2. VS2015. Need to install all the components? or can only individual?
Problem solved! The error was in the memory module on the motherboard. Thanks for answers! Last question: what edition of VS do you recommend for compiling a tensor flow?