TensorFlow version of a CPU and GPU versions of the points, CPU version installation is relatively simple, according to official documents TensorFlow be installed. But the CPU version of the CPU can only be used to calculate the low computational efficiency. For simple model calculation can use CPU mode, but for complex models require training on the GPU support.
First, the basic environmental requirements:
System: windows, Mac, Linux can
Graphics: NVIDIA graphics cards generally
Second, the installation environment:
1, the installation Anconda:
Anaconda Python is supported by the leading open platform for scientific data. Anaconda is a high-performance version of the open source distributed version of Python and R, including more than 100 of the most popular Python for scientific data, R and Scala software package. And supports a variety of GPU acceleration key installation package, such tensorflow gpu Edition.
Choose their own corresponding system, it is recommended Python3.7 version of
the official website: click to open
2, CUDA environment
CUDA (Compute Unified Device Architecture), is the introduction of the NVIDIA graphics card makers computing platform. CUDA ™ is a launch of NVIDIA's general purpose parallel computing architecture that enable GPU to solve complex computational problems. It contains CUDA instruction set architecture (ISA) and the interior of the GPU parallel computing engine.
Tensorflow GPU CUDA is the foundation
looks into its computer graphics for what version of CUDA (usually CUDA10)
to view Windows system
official
3, download, install CUDA
download the corresponding system corresponds to your graphics adapted version of CUDA
Official Download
Windows download Graphical of easy to install
Linux installation tutorial
Third, the installation Tensorflow GPU:
Anconda configuration, see: Click to open the
Open Anconda graphical interface:
Check Tensorflow GPU
right-click on the selected version (1.13 or 1.14 is recommended)
and then click Apply to install
the installation can enjoy Tensorflow GPU version after the completion of
the command line:
conda install tensorflow-gpu==1.14
