Environment Windows 10 Visual Studio 2017 (may not be required) Anaconda 3 (Python 3.7.4)
Installation of various software Be careful about the version of each package As of January 31, 2020, the tensorflow page states: https://www.tensorflow.org/install/gpu
Software requirements The following NVIDIA® software must be installed on your system:
NVIDIA® GPU drivers —CUDA 10.1 requires 418.x or higher.
Installed is ver.441.87 CUDA® Toolkit —TensorFlow supports CUDA 10.1 (TensorFlow >= 2.1.0) ※https://developer.nvidia.com/cuda-toolkit-archive CUPTI ships with the CUDA Toolkit.
Included in CUDA Toolkit cuDNN SDK (>= 7.6) (Optional) TensorRT 6.0 to improve latency and throughput for inference on some models.
Visual Studio Redistribution v14.0 is required (?), And Visual Studio 2017 was originally included, but just in case, the runtime of Visual Studio 2015, 2017, 2019 is installed. https://support.microsoft.com/en-us/help/2977003/the-latest-supported-visual-c-downloads
Post-installation PATH settings C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.1\bin C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.1\extras\CUPTI\lib64 C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.1\include C:\SoftTool\cuDNN\cuda\bin
Pip package See the pip install guide for available packages, systems requirements, and instructions. To pip install a TensorFlow package with GPU support, choose a stable or development package:
pip install tensorflow-gpu
Confirmation command reference https://nigimitama.hatenablog.jp/entry/2018/12/24/022756
Recommended Posts