search

bdzyubak / tensorflow-sandbox

A repository for studying applications of Deep Learning across fields, and demonstrating samples of my code and project managment
0 stars 0 forks source link

Enable use of GPUs #25

Open bdzyubak opened 2 years ago

bdzyubak commented 2 years ago

Training on a desktop machine is fairly slow using CPUs only. However, many machines having gaming GPUs (!), so the automated initializers in this project should be enabled to use a GPU if one is available.

bdzyubak commented 2 years ago

The GPU setup involves the following steps: `Follow this tutorial Tensorflow GPU I did it and it works perfect.

Attention! - install version 9.0! newer version is not supported by Tensorflow-gpu

Steps:

1) Uninstall your old tensorflow 2) Install tensorflow-gpu pip install tensorflow-gpu 3) Install Nvidia Graphics Card & Drivers (you probably already have) 4) Download & Install CUDA 5) Download & Install cuDNN - Requires free registration for the NVIDIA developer program 6) Verify by simple program 7) from tensorflow.python.client import device_lib print(device_lib.list_local_devices()) or tf.config.list_physical_devices()` https://stackoverflow.com/questions/51306862/how-do-i-use-tensorflow-gpu

Comments: 1) requirements.txt will need to be updated to include tensoflow-gpu. 2) README will need to include the CUDA/cuDNN steps as a prerequsite. 3) The command prompt and code need to be closed before running the list devices command.

bdzyubak commented 2 years ago

GUIDES: CUDA - https://docs.nvidia.com/cuda/cuda-installation-guide-microsoft-windows/index.html cuDNN - https://docs.nvidia.com/deeplearning/cudnn/install-guide/index.html#install-windows (Note that this also requires zlib used by cuDNN)